Battam commited on
Commit
ef2fda3
·
verified ·
1 Parent(s): 2f8d870

switch to the typed pointer tables (data/ and meta/)

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. README.md +516 -6
  2. cards/ansible-galaxy-collections.yaml +231 -0
  3. cards/apache-jira.yaml +295 -0
  4. cards/arquivo-pt.yaml +354 -0
  5. cards/artifact-hub-helm-charts.yaml +259 -0
  6. cards/bioinformatics-public-archives.yaml +305 -0
  7. cards/centos-cbs-build-logs.yaml +241 -0
  8. cards/chromium-luci-builds.yaml +244 -0
  9. cards/common-crawl.yaml +1344 -0
  10. cards/conda-forge-packages.yaml +260 -0
  11. cards/cran.yaml +248 -0
  12. cards/crates-io.yaml +489 -0
  13. cards/ctftime.yaml +217 -0
  14. cards/datacite-dois.yaml +308 -0
  15. cards/debian-bugs.yaml +1123 -0
  16. cards/debian-sources.yaml +235 -0
  17. cards/debian-ubuntu-binary-packages.yaml +265 -0
  18. cards/discourse-forums.yaml +284 -0
  19. cards/docker-hub.yaml +268 -0
  20. cards/egov-japan-data.yaml +373 -0
  21. cards/end-of-term-web-archive.yaml +302 -0
  22. cards/enron-spreadsheets-emails.yaml +220 -0
  23. cards/eu-cellar-documents.yaml +315 -0
  24. cards/european-parliament-plenary.yaml +270 -0
  25. cards/f-droid-apps.yaml +302 -0
  26. cards/fatcat-files.yaml +213 -0
  27. cards/fedora-koji-build-logs.yaml +234 -0
  28. cards/freebsd-bugzilla.yaml +321 -0
  29. cards/fuse-spreadsheets-common-crawl.yaml +237 -0
  30. cards/fuse-spreadsheets.yaml +216 -0
  31. cards/gcc-bugzilla.yaml +286 -0
  32. cards/go-module-proxy-zips.yaml +461 -0
  33. cards/govdocs1-files.yaml +1366 -0
  34. cards/govdocs1.yaml +220 -0
  35. cards/govinfo-bulk-data.yaml +264 -0
  36. cards/hackage-packages.yaml +239 -0
  37. cards/hacker-news-threads.yaml +234 -0
  38. cards/harvard-dataverse.yaml +293 -0
  39. cards/hyperkitty-mailing-list-archives.yaml +211 -0
  40. cards/if-archive.yaml +292 -0
  41. cards/internet-archive-open-licensed.yaml +319 -0
  42. cards/kde-webkit-gentoo-bugzilla.yaml +331 -0
  43. cards/kernel-bugzilla.yaml +318 -0
  44. cards/launchpad-ubuntu-bugs.yaml +289 -0
  45. cards/libreoffice-bugzilla.yaml +323 -0
  46. cards/lichess.yaml +497 -0
  47. cards/linux-kernel-mail-archives.yaml +242 -0
  48. cards/loghub.yaml +460 -0
  49. cards/mastodon-public-timelines.yaml +296 -0
  50. cards/mendeley-data.yaml +313 -0
README.md CHANGED
@@ -2,24 +2,449 @@
2
  license: cc0-1.0
3
  pretty_name: Agentic Env Index
4
  size_categories:
5
- - 1K<n<10K
6
  tags:
7
  - data-catalog
8
  - manifest
9
  - urls
10
  - provenance
11
  configs:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
12
  - config_name: enron-spreadsheets-emails
13
- default: true
14
  data_files:
15
  - split: train
16
  path:
17
  - "data/enron-spreadsheets-emails/*/*.parquet"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
18
  - config_name: lichess
19
  data_files:
20
  - split: train
21
  path:
22
  - "data/lichess/*/*.parquet"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
23
  ---
24
  <!-- artifact-pointer-layout: v2 -->
25
 
@@ -27,7 +452,7 @@ configs:
27
 
28
  An index of materials for building agent environments: not a list of environments, and no content is copied here.
29
 
30
- It covers 1,383 public files and records (2.64 TB) from 2 sources. Each row says where one item can be downloaded, how large it is, which hashes verify it and under which license it is offered. The files themselves stay at their sources.
31
 
32
  ## Use
33
 
@@ -39,18 +464,103 @@ SELECT source, uri, size_bytes, lower(hex(sha256)) AS sha256 FROM 'hf://datasets
39
  WHERE ext = 'pdf' ORDER BY size_bytes DESC LIMIT 10;
40
  ```
41
 
42
- 3. Download and verify: `python tools/fetch.py --help` (filters by tag, type, source, license tier and size; checks every hash). See [docs/selecting.md](docs/selecting.md).
43
-
44
  ## Sources
45
 
46
  | Source | Rows | Size | Top types | Training | Redistribution | Unit | Link check |
47
  |---|---:|---:|---|---|---|---|---|
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
48
  | Enron spreadsheets and emails (`enron-spreadsheets-emails`) | 14 | 9.85 GB | 7z | yes | yes | file | 2026-10-01: 100% of 14 reachable |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
49
  | Lichess open database (`lichess`) | 1,369 | 2.63 TB | zst | yes | yes 94%, conditional 6% | file | 2026-10-01: 100% of 1,369 reachable |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
50
 
51
  The training and redistribution tiers are the source's license tiers; rows can carry their own tier in the license columns. A row is one file unless the unit says otherwise.
52
 
53
- More: [docs/selecting.md](docs/selecting.md) (fields, the default filter and examples); [meta/sources.json](meta/sources.json) (per-source details); [meta/coverage.json](meta/coverage.json) (totals by license tier, source, unit and size).
 
 
54
 
55
  ## License
56
 
 
2
  license: cc0-1.0
3
  pretty_name: Agentic Env Index
4
  size_categories:
5
+ - 100M<n<1B
6
  tags:
7
  - data-catalog
8
  - manifest
9
  - urls
10
  - provenance
11
  configs:
12
+ - config_name: ansible-galaxy-collections
13
+ data_files:
14
+ - split: train
15
+ path:
16
+ - "data/ansible-galaxy-collections/*/*.parquet"
17
+ - config_name: apache-jira
18
+ data_files:
19
+ - split: train
20
+ path:
21
+ - "data/apache-jira/*/*.parquet"
22
+ - config_name: arquivo-pt
23
+ data_files:
24
+ - split: train
25
+ path:
26
+ - "data/arquivo-pt/*/*.parquet"
27
+ - config_name: artifact-hub-helm-charts
28
+ data_files:
29
+ - split: train
30
+ path:
31
+ - "data/artifact-hub-helm-charts/*/*.parquet"
32
+ - config_name: arxiv-source
33
+ data_files:
34
+ - split: train
35
+ path:
36
+ - "data/arxiv-source/*/*.parquet"
37
+ - config_name: bioinformatics-public-archives
38
+ data_files:
39
+ - split: train
40
+ path:
41
+ - "data/bioinformatics-public-archives/*/*.parquet"
42
+ - config_name: centos-cbs-build-logs
43
+ data_files:
44
+ - split: train
45
+ path:
46
+ - "data/centos-cbs-build-logs/*/*.parquet"
47
+ - config_name: chromium-luci-builds
48
+ data_files:
49
+ - split: train
50
+ path:
51
+ - "data/chromium-luci-builds/*/*.parquet"
52
+ - config_name: common-crawl
53
+ data_files:
54
+ - split: train
55
+ path:
56
+ - "data/common-crawl/*/*.parquet"
57
+ - config_name: conda-forge-packages
58
+ data_files:
59
+ - split: train
60
+ path:
61
+ - "data/conda-forge-packages/*/*.parquet"
62
+ - config_name: cpan-distributions
63
+ data_files:
64
+ - split: train
65
+ path:
66
+ - "data/cpan-distributions/*/*.parquet"
67
+ - config_name: cran
68
+ data_files:
69
+ - split: train
70
+ path:
71
+ - "data/cran/*/*.parquet"
72
+ - config_name: crates-io
73
+ data_files:
74
+ - split: train
75
+ path:
76
+ - "data/crates-io/*/*.parquet"
77
+ - config_name: ctftime
78
+ data_files:
79
+ - split: train
80
+ path:
81
+ - "data/ctftime/*/*.parquet"
82
+ - config_name: datacite-dois
83
+ data_files:
84
+ - split: train
85
+ path:
86
+ - "data/datacite-dois/*/*.parquet"
87
+ - config_name: debian-bugs
88
+ data_files:
89
+ - split: train
90
+ path:
91
+ - "data/debian-bugs/*/*.parquet"
92
+ - config_name: debian-sources
93
+ data_files:
94
+ - split: train
95
+ path:
96
+ - "data/debian-sources/*/*.parquet"
97
+ - config_name: debian-ubuntu-binary-packages
98
+ data_files:
99
+ - split: train
100
+ path:
101
+ - "data/debian-ubuntu-binary-packages/*/*.parquet"
102
+ - config_name: discourse-forums
103
+ data_files:
104
+ - split: train
105
+ path:
106
+ - "data/discourse-forums/*/*.parquet"
107
+ - config_name: docker-hub
108
+ data_files:
109
+ - split: train
110
+ path:
111
+ - "data/docker-hub/*/*.parquet"
112
+ - config_name: egov-japan-data
113
+ data_files:
114
+ - split: train
115
+ path:
116
+ - "data/egov-japan-data/*/*.parquet"
117
+ - config_name: end-of-term-web-archive
118
+ data_files:
119
+ - split: train
120
+ path:
121
+ - "data/end-of-term-web-archive/*/*.parquet"
122
  - config_name: enron-spreadsheets-emails
 
123
  data_files:
124
  - split: train
125
  path:
126
  - "data/enron-spreadsheets-emails/*/*.parquet"
127
+ - config_name: eu-cellar-documents
128
+ data_files:
129
+ - split: train
130
+ path:
131
+ - "data/eu-cellar-documents/*/*.parquet"
132
+ - config_name: european-parliament-plenary
133
+ data_files:
134
+ - split: train
135
+ path:
136
+ - "data/european-parliament-plenary/*/*.parquet"
137
+ - config_name: f-droid-apps
138
+ data_files:
139
+ - split: train
140
+ path:
141
+ - "data/f-droid-apps/*/*.parquet"
142
+ - config_name: fatcat-files
143
+ data_files:
144
+ - split: train
145
+ path:
146
+ - "data/fatcat-files/*/*.parquet"
147
+ - config_name: fedora-koji-build-logs
148
+ data_files:
149
+ - split: train
150
+ path:
151
+ - "data/fedora-koji-build-logs/*/*.parquet"
152
+ - config_name: freebsd-bugzilla
153
+ data_files:
154
+ - split: train
155
+ path:
156
+ - "data/freebsd-bugzilla/*/*.parquet"
157
+ - config_name: fuse-spreadsheets
158
+ default: true
159
+ data_files:
160
+ - split: train
161
+ path:
162
+ - "data/fuse-spreadsheets/*/*.parquet"
163
+ - config_name: fuse-spreadsheets-common-crawl
164
+ data_files:
165
+ - split: train
166
+ path:
167
+ - "data/fuse-spreadsheets-common-crawl/*/*.parquet"
168
+ - config_name: gcc-bugzilla
169
+ data_files:
170
+ - split: train
171
+ path:
172
+ - "data/gcc-bugzilla/*/*.parquet"
173
+ - config_name: gh-archive-hourly
174
+ data_files:
175
+ - split: train
176
+ path:
177
+ - "data/gh-archive-hourly/*/*.parquet"
178
+ - config_name: github-gap-language-repos
179
+ data_files:
180
+ - split: train
181
+ path:
182
+ - "data/github-gap-language-repos/*/*.parquet"
183
+ - config_name: github-release-assets
184
+ data_files:
185
+ - split: train
186
+ path:
187
+ - "data/github-release-assets/*/*.parquet"
188
+ - config_name: go-module-proxy-zips
189
+ data_files:
190
+ - split: train
191
+ path:
192
+ - "data/go-module-proxy-zips/*/*.parquet"
193
+ - config_name: govdocs1
194
+ data_files:
195
+ - split: train
196
+ path:
197
+ - "data/govdocs1/*/*.parquet"
198
+ - config_name: govdocs1-files
199
+ data_files:
200
+ - split: train
201
+ path:
202
+ - "data/govdocs1-files/*/*.parquet"
203
+ - config_name: govinfo-bulk-data
204
+ data_files:
205
+ - split: train
206
+ path:
207
+ - "data/govinfo-bulk-data/*/*.parquet"
208
+ - config_name: hackage-packages
209
+ data_files:
210
+ - split: train
211
+ path:
212
+ - "data/hackage-packages/*/*.parquet"
213
+ - config_name: hacker-news-threads
214
+ data_files:
215
+ - split: train
216
+ path:
217
+ - "data/hacker-news-threads/*/*.parquet"
218
+ - config_name: harvard-dataverse
219
+ data_files:
220
+ - split: train
221
+ path:
222
+ - "data/harvard-dataverse/*/*.parquet"
223
+ - config_name: homebrew-packages
224
+ data_files:
225
+ - split: train
226
+ path:
227
+ - "data/homebrew-packages/*/*.parquet"
228
+ - config_name: hyperkitty-mailing-list-archives
229
+ data_files:
230
+ - split: train
231
+ path:
232
+ - "data/hyperkitty-mailing-list-archives/*/*.parquet"
233
+ - config_name: if-archive
234
+ data_files:
235
+ - split: train
236
+ path:
237
+ - "data/if-archive/*/*.parquet"
238
+ - config_name: internet-archive-open-licensed
239
+ data_files:
240
+ - split: train
241
+ path:
242
+ - "data/internet-archive-open-licensed/*/*.parquet"
243
+ - config_name: kde-webkit-gentoo-bugzilla
244
+ data_files:
245
+ - split: train
246
+ path:
247
+ - "data/kde-webkit-gentoo-bugzilla/*/*.parquet"
248
+ - config_name: kernel-bugzilla
249
+ data_files:
250
+ - split: train
251
+ path:
252
+ - "data/kernel-bugzilla/*/*.parquet"
253
+ - config_name: launchpad-ubuntu-bugs
254
+ data_files:
255
+ - split: train
256
+ path:
257
+ - "data/launchpad-ubuntu-bugs/*/*.parquet"
258
+ - config_name: lean-toolchains-and-packages
259
+ data_files:
260
+ - split: train
261
+ path:
262
+ - "data/lean-toolchains-and-packages/*/*.parquet"
263
+ - config_name: libreoffice-bugzilla
264
+ data_files:
265
+ - split: train
266
+ path:
267
+ - "data/libreoffice-bugzilla/*/*.parquet"
268
  - config_name: lichess
269
  data_files:
270
  - split: train
271
  path:
272
  - "data/lichess/*/*.parquet"
273
+ - config_name: linux-kernel-mail-archives
274
+ data_files:
275
+ - split: train
276
+ path:
277
+ - "data/linux-kernel-mail-archives/*/*.parquet"
278
+ - config_name: loghub
279
+ data_files:
280
+ - split: train
281
+ path:
282
+ - "data/loghub/*/*.parquet"
283
+ - config_name: mastodon-public-timelines
284
+ data_files:
285
+ - split: train
286
+ path:
287
+ - "data/mastodon-public-timelines/*/*.parquet"
288
+ - config_name: mendeley-data
289
+ data_files:
290
+ - split: train
291
+ path:
292
+ - "data/mendeley-data/*/*.parquet"
293
+ - config_name: modelscope
294
+ data_files:
295
+ - split: train
296
+ path:
297
+ - "data/modelscope/*/*.parquet"
298
+ - config_name: mozilla-bugzilla
299
+ data_files:
300
+ - split: train
301
+ path:
302
+ - "data/mozilla-bugzilla/*/*.parquet"
303
+ - config_name: mozilla-treeherder-ci-logs
304
+ data_files:
305
+ - split: train
306
+ path:
307
+ - "data/mozilla-treeherder-ci-logs/*/*.parquet"
308
+ - config_name: napierone-digital-corpora
309
+ data_files:
310
+ - split: train
311
+ path:
312
+ - "data/napierone-digital-corpora/*/*.parquet"
313
+ - config_name: napierone-files
314
+ data_files:
315
+ - split: train
316
+ path:
317
+ - "data/napierone-files/*/*.parquet"
318
+ - config_name: npm
319
+ data_files:
320
+ - split: train
321
+ path:
322
+ - "data/npm/*/*.parquet"
323
+ - config_name: nuget
324
+ data_files:
325
+ - split: train
326
+ path:
327
+ - "data/nuget/*/*.parquet"
328
+ - config_name: nvd-cve
329
+ data_files:
330
+ - split: train
331
+ path:
332
+ - "data/nvd-cve/*/*.parquet"
333
+ - config_name: opendev-zuul-ci
334
+ data_files:
335
+ - split: train
336
+ path:
337
+ - "data/opendev-zuul-ci/*/*.parquet"
338
+ - config_name: openstreetmap-planet
339
+ data_files:
340
+ - split: train
341
+ path:
342
+ - "data/openstreetmap-planet/*/*.parquet"
343
+ - config_name: osv-vulnerabilities
344
+ data_files:
345
+ - split: train
346
+ path:
347
+ - "data/osv-vulnerabilities/*/*.parquet"
348
+ - config_name: pmc-open-access
349
+ data_files:
350
+ - split: train
351
+ path:
352
+ - "data/pmc-open-access/*/*.parquet"
353
+ - config_name: project-gutenberg
354
+ data_files:
355
+ - split: train
356
+ path:
357
+ - "data/project-gutenberg/*/*.parquet"
358
+ - config_name: pypi
359
+ data_files:
360
+ - split: train
361
+ path:
362
+ - "data/pypi/*/*.parquet"
363
+ - config_name: reddit-threads-arctic-shift
364
+ data_files:
365
+ - split: train
366
+ path:
367
+ - "data/reddit-threads-arctic-shift/*/*.parquet"
368
+ - config_name: redhat-bugzilla
369
+ data_files:
370
+ - split: train
371
+ path:
372
+ - "data/redhat-bugzilla/*/*.parquet"
373
+ - config_name: rosetta-code-github-mirror
374
+ data_files:
375
+ - split: train
376
+ path:
377
+ - "data/rosetta-code-github-mirror/*/*.parquet"
378
+ - config_name: rubygems
379
+ data_files:
380
+ - split: train
381
+ path:
382
+ - "data/rubygems/*/*.parquet"
383
+ - config_name: safedocs-cc-main-2021-31
384
+ data_files:
385
+ - split: train
386
+ path:
387
+ - "data/safedocs-cc-main-2021-31/*/*.parquet"
388
+ - config_name: sec-edgar-filings
389
+ data_files:
390
+ - split: train
391
+ path:
392
+ - "data/sec-edgar-filings/*/*.parquet"
393
+ - config_name: smithsonian-open-access
394
+ data_files:
395
+ - split: train
396
+ path:
397
+ - "data/smithsonian-open-access/*/*.parquet"
398
+ - config_name: solver-benchmark-instances
399
+ data_files:
400
+ - split: train
401
+ path:
402
+ - "data/solver-benchmark-instances/*/*.parquet"
403
+ - config_name: sourceware-public-inbox
404
+ data_files:
405
+ - split: train
406
+ path:
407
+ - "data/sourceware-public-inbox/*/*.parquet"
408
+ - config_name: stackexchange-api-posts
409
+ data_files:
410
+ - split: train
411
+ path:
412
+ - "data/stackexchange-api-posts/*/*.parquet"
413
+ - config_name: stackexchange-dump
414
+ data_files:
415
+ - split: train
416
+ path:
417
+ - "data/stackexchange-dump/*/*.parquet"
418
+ - config_name: ubuntu-irc-logs
419
+ data_files:
420
+ - split: train
421
+ path:
422
+ - "data/ubuntu-irc-logs/*/*.parquet"
423
+ - config_name: uk-parliament-hansard
424
+ data_files:
425
+ - split: train
426
+ path:
427
+ - "data/uk-parliament-hansard/*/*.parquet"
428
+ - config_name: wikimedia-talk-dumps
429
+ data_files:
430
+ - split: train
431
+ path:
432
+ - "data/wikimedia-talk-dumps/*/*.parquet"
433
+ - config_name: worldbank-open-data
434
+ data_files:
435
+ - split: train
436
+ path:
437
+ - "data/worldbank-open-data/*/*.parquet"
438
+ - config_name: zenodo10k-slides
439
+ data_files:
440
+ - split: train
441
+ path:
442
+ - "data/zenodo10k-slides/*/*.parquet"
443
+ - config_name: zulip-public-chats
444
+ data_files:
445
+ - split: train
446
+ path:
447
+ - "data/zulip-public-chats/*/*.parquet"
448
  ---
449
  <!-- artifact-pointer-layout: v2 -->
450
 
 
452
 
453
  An index of materials for building agent environments: not a list of environments, and no content is copied here.
454
 
455
+ It covers 885,728,417 public files and records (3.91 PB) from 87 sources. Each row says where one item can be downloaded, how large it is, which hashes verify it and under which license it is offered. The files themselves stay at their sources.
456
 
457
  ## Use
458
 
 
464
  WHERE ext = 'pdf' ORDER BY size_bytes DESC LIMIT 10;
465
  ```
466
 
 
 
467
  ## Sources
468
 
469
  | Source | Rows | Size | Top types | Training | Redistribution | Unit | Link check |
470
  |---|---:|---:|---|---|---|---|---|
471
+ | Ansible Galaxy collections (`ansible-galaxy-collections`) | 52,784 | 0 B (52,784 unsized) | gz | yes 70%, unknown 28%, no 2%, conditional 0% | yes 36%, conditional 35%, unknown 29% | file | not checked |
472
+ | Apache Jira issues (`apache-jira`) | 1,179,062 | 1.28 GB | txt | unknown | unknown | API response | not checked |
473
+ | Arquivo.pt office documents (`arquivo-pt`) | 85,980 | 3.45 GB (82,217 unsized) | pdf, doc, ppt | unknown | no | file | not checked |
474
+ | Artifact Hub Helm charts (`artifact-hub-helm-charts`) | 127,281 | 0 B (127,281 unsized) | tgz | unknown 82%, yes 18% | unknown 82%, yes 16%, conditional 2% | file | not checked |
475
+ | arXiv source packages (`arxiv-source`) | 3,165,721 | 14.1 TB | gz, pdf, tar | unknown 75%, yes 20%, conditional 5% | no 60%, yes 19%, unknown 14%, conditional 6% | archive member | not checked |
476
+ | Bioinformatics public data archives (`bioinformatics-public-archives`) | 278,202 | 7.81 TB (237,457 unsized) | gz, (none), zip | yes 96%, unknown 4% | yes 96%, unknown 4%, conditional 0% | file | not checked |
477
+ | CentOS Community Build System build logs (`centos-cbs-build-logs`) | 113,676 | 155 GB | log | unknown | unknown | file | not checked |
478
+ | Chromium LUCI failed build logs (`chromium-luci-builds`) | 8,665,027 | 0 B (8,665,027 unsized) | log | unknown | unknown | file | not checked |
479
+ | Common Crawl office files (`common-crawl`) | 148,118,699 | 119 TB | pdf, txt, xml | unknown | unknown | WARC record | not checked |
480
+ | conda-forge packages (`conda-forge-packages`) | 3,369,245 | 25.4 TB | conda, bz2 | yes 94%, unknown 6% | yes 73%, conditional 22%, unknown 6% | file | not checked |
481
+ | CPAN distributions (`cpan-distributions`) | 165,740 | 38.4 GB | gz, tgz, zip | yes 71%, unknown 29% | conditional 64%, unknown 29%, yes 7% | file | not checked |
482
+ | CRAN packages (`cran`) | 25,299 | 18.9 GB | gz | yes | conditional 67%, yes 33% | file | 2026-10-02: 100% of 25,284 reachable |
483
+ | crates.io packages (`crates-io`) | 3,252,145 | 663 GB | crate | yes 96%, unknown 3%, conditional 1% | yes 87%, conditional 9%, unknown 3% | file | not checked |
484
+ | CTFtime events, tasks and writeups (`ctftime`) | 75,162 | 489 MB (33,023 unsized) | html, json | no | unknown | file | not checked |
485
+ | DataCite DOI metadata (`datacite-dois`) | 31,612,027 | 2.23 TB (31,609,847 unsized) | json, (none), zip | yes 99%, unknown 1%, conditional 0% | yes 99%, unknown 1%, conditional 0% | API response | not checked |
486
+ | Debian bug reports (`debian-bugs`) | 2,244,932 | 145 GB | log, summary | unknown | unknown | file | 2026-10-01: 97% of 60 reachable |
487
+ | Debian source packages (`debian-sources`) | 115,614 | 99.2 GB | xz, dsc, gz | yes | conditional | file | not checked |
488
+ | Debian and Ubuntu binary packages (`debian-ubuntu-binary-packages`) | 676,230 | 2.11 TB | deb, xz | yes 93%, unknown 7% | conditional 93%, unknown 7% | file | not checked |
489
+ | Discourse forum posts (`discourse-forums`) | 12,023,523 | 8.74 GB | md | unknown 52%, conditional 38%, yes 9% | unknown 53%, conditional 44%, yes 3% | API response | not checked |
490
+ | Docker Hub container images (`docker-hub`) | 593,368 | 221 TB | json | conditional | unknown | file | not checked |
491
+ | e-Gov Data Portal (Japan) (`egov-japan-data`) | 292,050 | 15.4 GB (244,867 unsized) | html, do, xls | yes 99%, unknown 1%, conditional 0% | yes 99%, unknown 1%, conditional 0% | file | not checked |
492
+ | End of Term Web Archive (`end-of-term-web-archive`) | 337,638,097 | 149 TB | xml, pdf, txt | yes | yes | WARC record | 2026-10-02: 100% of 180 reachable |
493
  | Enron spreadsheets and emails (`enron-spreadsheets-emails`) | 14 | 9.85 GB | 7z | yes | yes | file | 2026-10-01: 100% of 14 reachable |
494
+ | EU Publications Office CELLAR documents (`eu-cellar-documents`) | 1,921,815 | 803 GB (1,396,072 unsized) | pdf, xml, html | yes | yes | file | not checked |
495
+ | European Parliament plenary records (`european-parliament-plenary`) | 1,233,677 | 38.0 GB (712,457 unsized) | pdf, xml, doc | yes | yes | file | not checked |
496
+ | F-Droid apps (`f-droid-apps`) | 196,376 | 7.30 TB | apk, gz, zip | yes 100%, unknown 0% | conditional 71%, yes 28%, unknown 0% | file | not checked |
497
+ | fatcat scholarly files (`fatcat-files`) | 47,163,157 | 70.1 TB (27,749 unsized) | pdf, (none), xml | unknown | unknown | file | not checked |
498
+ | Fedora Koji build logs (`fedora-koji-build-logs`) | 119,036 | 194 GB | log | unknown | unknown | file | not checked |
499
+ | FreeBSD Bugzilla bugs (`freebsd-bugzilla`) | 573,793 | 5.67 GB | json, diff, patch | unknown | unknown | API response | not checked |
500
+ | FUSE spreadsheets (`fuse-spreadsheets`) | 1 | 9.39 GB | zip | yes | yes | file | 2026-10-01: 100% of 1 reachable |
501
+ | FUSE spreadsheets (Common Crawl records) (`fuse-spreadsheets-common-crawl`) | 249,376 | 7.59 GB | xls, xlsx, xltx | unknown | unknown | WARC record | not checked |
502
+ | GCC Bugzilla bugs (`gcc-bugzilla`) | 186,889 | 4.93 GB (53 unsized) | json, patch, c | unknown | unknown | API response | not checked |
503
+ | GH Archive hourly event files (`gh-archive-hourly`) | 136,799 | 5.02 TB | gz | unknown | unknown | file | not checked |
504
+ | GitHub repositories (selected languages) (`github-gap-language-repos`) | 6,707,505 | 718 GB | scala, v, (none) | yes 55%, unknown 45% | unknown 45%, yes 43%, conditional 12% | file | not checked |
505
+ | GitHub release assets (`github-release-assets`) | 69,081 | 1.37 TB (18,739 unsized) | gz, png, zip | unknown 62%, yes 38% | unknown 62%, yes 29%, conditional 8% | file | not checked |
506
+ | Go module mirror zips (`go-module-proxy-zips`) | 59,120,216 | 0 B (59,120,216 unsized) | zip | yes 59%, unknown 41%, conditional 0% | yes 49%, unknown 41%, conditional 9% | file | not checked |
507
+ | GovDocs1 documents (`govdocs1`) | 1,000 | 333 GB | zip | yes | yes | file | 2026-10-01: 100% of 1,000 reachable |
508
+ | GovDocs1 files (`govdocs1-files`) | 986,279 | 501 GB | pdf, html, jpg | yes | yes | archive member | not checked |
509
+ | govinfo bulk data (`govinfo-bulk-data`) | 419,274 | 56.4 GB | xml, zip, html | yes | yes | file | 2026-10-01: 9% of 351 reachable |
510
+ | Hackage packages (`hackage-packages`) | 154,127 | 20.0 GB | gz | yes 93%, unknown 7% | yes 84%, conditional 9%, unknown 7% | file | not checked |
511
+ | Hacker News discussion threads (`hacker-news-threads`) | 4,790,481 | 0 B (4,790,481 unsized) | json | no | no | file | not checked |
512
+ | Harvard Dataverse (`harvard-dataverse`) | 835,340 | 56.0 TB | jpg, pdf, dcm | yes 76%, conditional 24% | yes 73%, conditional 27% | file | not checked |
513
+ | Homebrew packages (`homebrew-packages`) | 81,275 | 572 GB (31,254 unsized) | gz, rb, dmg | yes 87%, unknown 13% | yes 58%, conditional 29%, unknown 13% | file | not checked |
514
+ | Mailman 3 HyperKitty list archives (`hyperkitty-mailing-list-archives`) | 80,365 | 0 B (80,365 unsized) | gz | unknown | unknown | file | not checked |
515
+ | IF Archive text adventures (`if-archive`) | 23,819 | 53.3 GB | zip, png, txt | no 99%, unknown 1% | no 99%, unknown 1% | file | not checked |
516
+ | Internet Archive software and open media (`internet-archive-open-licensed`) | 3,350,089 | 160 TB | sig, zst, xz | unknown 98%, yes 1%, conditional 0% | unknown 98%, yes 1%, conditional 0% | file | not checked |
517
+ | KDE, WebKit and Gentoo Bugzilla bugs (`kde-webkit-gentoo-bugzilla`) | 1,672,112 | 72.1 GB | json, log, patch | unknown | unknown | API response | not checked |
518
+ | Linux kernel Bugzilla bugs (`kernel-bugzilla`) | 125,846 | 10.6 GB | json, txt, (none) | unknown | unknown | API response | not checked |
519
+ | Launchpad Ubuntu bugs (`launchpad-ubuntu-bugs`) | 63,380 | 42.4 MB (17,013 unsized) | json, (none), log | unknown | unknown | API response | not checked |
520
+ | Lean toolchains and packages (`lean-toolchains-and-packages`) | 37,933 | 7.41 TB (872 unsized) | gz, zip, zst | yes 56%, unknown 44% | yes 56%, unknown 44%, conditional 0% | file | not checked |
521
+ | LibreOffice Bugzilla bugs (`libreoffice-bugzilla`) | 244,962 | 51.1 GB | json, png, odt | yes | conditional | API response | not checked |
522
  | Lichess open database (`lichess`) | 1,369 | 2.63 TB | zst | yes | yes 94%, conditional 6% | file | 2026-10-01: 100% of 1,369 reachable |
523
+ | Linux kernel mailing list archives (`linux-kernel-mail-archives`) | 405 | 0 B (405 unsized) | git | no | unknown | file | 2026-10-01: 0% of 405 reachable |
524
+ | Loghub system logs (`loghub`) | 19 | 6.00 GB | gz, zip | conditional | conditional | file | 2026-10-01: 100% of 19 reachable |
525
+ | Mastodon public posts (`mastodon-public-timelines`) | 4,914,927 | 1.73 GB (1,144,744 unsized) | html, png, jpg | unknown 61%, no 39% | unknown | API response | not checked |
526
+ | Mendeley Data research data files (`mendeley-data`) | 130,156 | 2.11 TB | tif, jpg, txt | yes 93%, conditional 6%, unknown 0% | yes 93%, conditional 7%, no 1%, unknown 0% | file | not checked |
527
+ | ModelScope datasets and models (`modelscope`) | 8,726,153 | 2.03 PB | mp4, jpg, parquet | yes 85%, unknown 10%, conditional 5% | yes 83%, unknown 10%, conditional 7%, no 0% | file | not checked |
528
+ | Mozilla Bugzilla bugs (`mozilla-bugzilla`) | 3,896,234 | 266 GB | json, txt, patch | unknown | unknown | file | not checked |
529
+ | Mozilla Treeherder failed job logs (`mozilla-treeherder-ci-logs`) | 37,531 | 80.4 GB (8,609 unsized) | log | unknown | unknown | file | not checked |
530
+ | NapierOne and Digital Corpora scenarios (`napierone-digital-corpora`) | 1,056 | 2.49 TB | zip, e01, xml | yes 97%, unknown 3% | conditional 97%, unknown 3% | file | 2026-10-01: 99% of 307 reachable |
531
+ | NapierOne files (`napierone-files`) | 659,576 | 922 GB | webp, jpg, 7z | yes 93%, unknown 7% | conditional 93%, unknown 7% | archive member | not checked |
532
+ | npm packages (`npm`) | 50,313,313 | 0 B (50,313,313 unsized) | tgz | yes 80%, unknown 19%, conditional 0% | yes 75%, unknown 19%, conditional 6% | file | not checked |
533
+ | NuGet packages (`nuget`) | 562,281 | 424 GB | nupkg | unknown 100%, yes 0% | unknown 100%, yes 0% | file | not checked |
534
+ | NVD CVE vulnerability records (`nvd-cve`) | 381,690 | 1.95 GB | json | yes | yes | API response | not checked |
535
+ | OpenDev Zuul CI job logs (`opendev-zuul-ci`) | 27,332 | 10.6 GB | gz | unknown | unknown | file | not checked |
536
+ | OpenStreetMap planet files (`openstreetmap-planet`) | 27,295,635 | 9.13 TB (14,266,744 unsized) | gz, txt, torrent | yes 100%, unknown 0% | conditional 100%, unknown 0% | file | not checked |
537
+ | OSV open source vulnerability database (`osv-vulnerabilities`) | 2,616,111 | 22.5 GB | json, csv, zip | unknown 82%, yes 18% | unknown 82%, yes 15%, conditional 3% | file | not checked |
538
+ | PubMed Central Open Access articles (`pmc-open-access`) | 4,318,404 | 1.23 TB | jpg, xml, txt | yes 61%, conditional 30%, unknown 9% | yes 61%, conditional 21%, unknown 9%, no 9% | file | not checked |
539
+ | Project Gutenberg ebooks (`project-gutenberg`) | 2,024,683 | 716 GB (942,233 unsized) | png, epub, mobi | yes 99%, unknown 1% | conditional 99%, no 1% | file | not checked |
540
+ | PyPI packages (`pypi`) | 45,486,331 | 129 TB | whl, gz, egg | yes 82%, unknown 17%, conditional 0% | yes 70%, unknown 17%, conditional 12% | file | not checked |
541
+ | Reddit threads (Arctic Shift archive) (`reddit-threads-arctic-shift`) | 318,856 | 0 B (318,856 unsized) | json | no | no | file | not checked |
542
+ | Red Hat Bugzilla bugs (`redhat-bugzilla`) | 1,198,320 | 83.5 GB | json, (none), log | unknown | unknown | API response | not checked |
543
+ | Rosetta Code (GitHub community mirror) (`rosetta-code-github-mirror`) | 148,991 | 153 MB | basic, j, py | unknown | unknown | file | not checked |
544
+ | RubyGems packages (`rubygems`) | 326,862 | 183 GB | gem | yes 72%, unknown 28%, conditional 0% | yes 69%, unknown 28%, conditional 3% | file | not checked |
545
+ | SafeDocs CC-MAIN-2021-31 PDF and UNSAFE (`safedocs-cc-main-2021-31`) | 13,457,931 | 14.9 TB | pdf, (none), pcap | unknown | unknown | byte range | not checked |
546
+ | SEC EDGAR company filings (`sec-edgar-filings`) | 18,788,783 | 0 B (18,788,783 unsized) | txt | yes | yes | file | not checked |
547
+ | Smithsonian Open Access (`smithsonian-open-access`) | 9,311,739 | 851 TB | jpg, tif, json | yes | yes | file | not checked |
548
+ | Solver benchmark instances (`solver-benchmark-instances`) | 108,083 | 145 GB | xz, i, yml | unknown 88%, yes 12% | unknown 88%, yes 12%, conditional 0% | file | not checked |
549
+ | Sourceware mailing list archives (`sourceware-public-inbox`) | 168 | 0 B (168 unsized) | git | unknown | unknown | file | not checked |
550
+ | Stack Exchange recent posts (API) (`stackexchange-api-posts`) | 3,084,502 | 5.03 GB | html | no | conditional | API response | not checked |
551
+ | Stack Exchange data dump (`stackexchange-dump`) | 3,629 | 960 GB | 7z | no 90%, yes 10% | no 90%, conditional 10% | file | 2026-10-01: 100% of 3,629 reachable |
552
+ | Ubuntu IRC logs (`ubuntu-irc-logs`) | 456,106 | 51.6 MB (229,171 unsized) | txt | conditional | conditional | file | 2026-10-02: 100% of 8,846 reachable |
553
+ | UK Parliament Hansard debates (`uk-parliament-hansard`) | 1,370,783 | 0 B (1,370,783 unsized) | json | yes | yes | API response | not checked |
554
+ | Wikimedia talk page dumps (`wikimedia-talk-dumps`) | 1,563 | 503 GB | bz2 | yes | conditional 97%, yes 3% | file | not checked |
555
+ | World Bank open data files (`worldbank-open-data`) | 33,928 | 334 GB (484 unsized) | zip, xlsx, csv | yes 99%, conditional 1%, unknown 0% | yes 98%, conditional 1%, unknown 0% | file | not checked |
556
+ | Zenodo10K slides (`zenodo10k-slides`) | 10,448 | 242 GB | pptx | yes 98%, conditional 1%, unknown 0% | yes 96%, conditional 4%, unknown 0% | file | 2026-10-01: 100% of 10,448 reachable |
557
+ | Zulip public chat archives (`zulip-public-chats`) | 1,300,638 | 239 MB | txt | no | conditional | API response | not checked |
558
 
559
  The training and redistribution tiers are the source's license tiers; rows can carry their own tier in the license columns. A row is one file unless the unit says otherwise.
560
 
561
+ Sources still being collected also have a `part-tail.parquet` in their batch folder: it holds the rows that do not yet fill a numbered part file and is replaced as collection goes on; once enough rows arrive they move into the next `part-<number>.parquet`, so a row's file and row number in the tail are not stable.
562
+
563
+ More: [meta/sources.json](meta/sources.json) (per-source details); [meta/coverage.json](meta/coverage.json) (totals by license tier, source, unit and size); [cards/](https://huggingface.co/datasets/Battam/agentic-env-index/tree/main/cards) (one card per source: what its files are, sampled and read by a person, with the evidence for each field).
564
 
565
  ## License
566
 
cards/ansible-galaxy-collections.yaml ADDED
@@ -0,0 +1,231 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: ansible-galaxy-collections
8
+ parent: null
9
+ title: "Ansible Galaxy"
10
+ description: "The tar.gz archive of every version of every collection on Ansible Galaxy (galaxy.ansible.com), the official content hub for Ansible configuration management; collections hold modules, roles, plugins and example playbooks. One pointer per collection version. The version list, sha256, sha512, license, dependencies and tags come from the Galaxy NG (pulp_ansible) content list, limited to the published repository; the deprecated flag comes from the collection index. Packages can contain malicious code; treat files as data and run nothing outside a sandbox."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per item (license field of galaxy.yml, a list of SPDX identifiers); the dominant tier follows from the counts"
20
+ licence_url: "https://galaxy.ansible.com"
21
+ licence_statement: "Each collection declares its own license; the site adds no license of its own."
22
+ per_collection: null
23
+ training_ok: conditional
24
+ redistribution_ok: conditional
25
+ attribution: null
26
+ per_file: "license_override.notes gives the license list of each version; license_override.spdx gives the SPDX expression (several identifiers joined with AND) where one is recognised. Versions whose namespace has opted out of AI training on Galaxy are marked training_ok no."
27
+ licence_tiers:
28
+ conditional/conditional: 9
29
+ no/conditional: 295
30
+ no/unknown: 355
31
+ no/yes: 268
32
+ unknown/unknown: 14888
33
+ yes/conditional: 18176
34
+ yes/yes: 18793
35
+ spdx_top:
36
+ "(none)": 15243
37
+ "0BSD": 23
38
+ AGPL-3.0-only: 313
39
+ AGPL-3.0-or-later: 359
40
+ Apache-2.0: 9058
41
+ "Apache-2.0 AND GPL-2.0-or-later AND MIT": 25
42
+ "Apache-2.0 AND GPL-3.0-or-later": 53
43
+ "Apache-2.0 AND MIT": 18
44
+ BSD-2-Clause: 205
45
+ BSD-3-Clause: 444
46
+ EPL-2.0: 434
47
+ EUPL-1.2: 83
48
+ GPL-2.0-only: 135
49
+ GPL-2.0-or-later: 7865
50
+ "GPL-2.0-or-later AND GPL-3.0-only AND GPL-3.0-or-later AND LGPL-3.0-only AND MIT AND BSD-3-Clause": 282
51
+ spdx_distinct: 55
52
+ licence_sample_check: null
53
+ unit:
54
+ pointer_level: null
55
+ pointer_count: 52784
56
+ units: {file: 52784}
57
+ state: null
58
+ expanded_count: null
59
+ bytes_total: 0
60
+ rows_without_size: 52784
61
+ collections:
62
+ distinct: 4566
63
+ top:
64
+ alvistack.ceph: 422
65
+ dim_3.chu_lille: 892
66
+ dim_3.example: 545
67
+ dim_3.feedim: 602
68
+ dim_3.infrastructure: 579
69
+ dim_3.nutrow: 811
70
+ redhatci.ocp: 789
71
+ waldur.marketplace: 1206
72
+ waldur.openstack: 1209
73
+ waldur.structure: 1209
74
+ sub_collections: null
75
+ batches: null
76
+ n_ext: 1
77
+ usable_unit: null
78
+ form:
79
+ by_ext: {gz: 52784}
80
+ bytes_p50: null
81
+ bytes_p95: null
82
+ bytes_max: null
83
+ size_quantiles: null
84
+ by_ext_size:
85
+ - {ext: gz, rows: 52784, bytes: 0, p50: null, p95: null, max: null}
86
+ over_cap:
87
+ cap: 61944279
88
+ rows_over: 0
89
+ bytes_over: 0
90
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
91
+ largest: []
92
+ date_years: null
93
+ media_type: {"(none)": 52784}
94
+ facets: null
95
+ artifact: null
96
+ source_origin: null
97
+ parse_cost: null
98
+ parse_quirks: null
99
+ structure:
100
+ self_describing: null
101
+ integrity_builtin: null
102
+ has_reference_output: null
103
+ invariants: null
104
+ tools: null
105
+ ships:
106
+ problem_statement: {present: null}
107
+ reference_solution: {share: null}
108
+ tests: {present: null}
109
+ env_definition: {present: null}
110
+ known_answer: {present: null}
111
+ generator: {present: null}
112
+ upstream_labels: {present: null}
113
+ class: null
114
+ subject:
115
+ domain_span: null
116
+ shape_span: null
117
+ skill_family_span: null
118
+ ont_domain: null
119
+ tech_vocabulary: null
120
+ risk:
121
+ contamination: none
122
+ overlap_check:
123
+ done: true
124
+ pointers_flagged: 0
125
+ pointers_removed: 0
126
+ by_class: {}
127
+ benches: {}
128
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
129
+ pii: null
130
+ pii_evidence: null
131
+ malware: null
132
+ ship_bytes: conditional
133
+ subset:
134
+ rule: "take all rows; licence, size and version are labels, not filters"
135
+ rows_in_subset: 52784
136
+ last_built: null
137
+ access:
138
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
139
+ verify: "Compare sha256 and sha512 (published). The size is not published (size_bytes null); record the measured size."
140
+ hosts: {"https://galaxy.ansible.com": 52784}
141
+ uri_example: "https://galaxy.ansible.com/api/v3/plugin/ansible/content/published/collections/artifacts/a10-acos_axapi-1.0.0.tar.gz"
142
+ needs_auth: {"false": 52784}
143
+ credential: null
144
+ hash_source: {source_published: 52784}
145
+ range_supported: null
146
+ reachable:
147
+ index_flag: {unknown: 52784}
148
+ index_check: null
149
+ our_sample: null
150
+ drift: null
151
+ sample: null
152
+ evidence:
153
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
154
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
155
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
156
+ class: {fill: unfilled, basis: "ships not judged"}
157
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
158
+ form.by_ext:
159
+ fill: auto
160
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
161
+ checked_at: "2026-10-03"
162
+ form.date_years:
163
+ fill: auto
164
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
165
+ checked_at: "2026-10-03"
166
+ form.over_cap:
167
+ fill: auto
168
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
169
+ checked_at: "2026-10-03"
170
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
171
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
172
+ form.size_quantiles:
173
+ fill: auto
174
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
175
+ checked_at: "2026-10-03"
176
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
177
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
178
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
179
+ provenance.licence_tiers:
180
+ fill: auto
181
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
182
+ checked_at: "2026-10-03"
183
+ provenance.spdx_top:
184
+ fill: auto
185
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
186
+ checked_at: "2026-10-03"
187
+ risk.contamination:
188
+ fill: auto
189
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
190
+ checked_at: "2026-10-03"
191
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
192
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
193
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
194
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
195
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
196
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
197
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
198
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
199
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ snapshot.rev:
208
+ fill: auto
209
+ basis: "last commit listed in the publish receipt of this source"
210
+ checked_at: "2026-10-03"
211
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
212
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
213
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
214
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ subset:
222
+ fill: auto
223
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
224
+ checked_at: "2026-10-03"
225
+ unit.pointer_count:
226
+ fill: auto
227
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
228
+ checked_at: "2026-10-03"
229
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
231
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/apache-jira.yaml ADDED
@@ -0,0 +1,295 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: apache-jira
8
+ parent: null
9
+ title: "Apache Jira issues"
10
+ description: "Issues of all public projects on Apache Jira (issues.apache.org). One pointer per issue; the hash covers the issue description (fields.description) as returned by the API, as UTF-8 bytes. Title, type, status and attachment counts are in source_meta."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: "1ba6d6c2095b87420afe27b56e210aae2c0f9882", at: "2026-10-01", rev_candidates: 8}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "No license stated for issue content"
17
+ licence_url: "https://issues.apache.org/jira"
18
+ licence_statement: "The issue tracker states no license for issue text."
19
+ per_collection: null
20
+ training_ok: unknown
21
+ redistribution_ok: unknown
22
+ attribution: null
23
+ per_file: null
24
+ licence_tiers: {unknown/unknown: 1179062}
25
+ spdx_top: {"(none)": 1179062}
26
+ spdx_distinct: 1
27
+ licence_sample_check: null
28
+ unit:
29
+ pointer_level: null
30
+ pointer_count: 1179062
31
+ units: {"API response": 1179062}
32
+ state: null
33
+ expanded_count: null
34
+ bytes_total: 1277864423
35
+ rows_without_size: 0
36
+ collections:
37
+ distinct: 661
38
+ top:
39
+ AMBARI: 26345
40
+ CAMEL: 25177
41
+ CASSANDRA: 21490
42
+ FLEX: 35395
43
+ FLINK: 40821
44
+ HBASE: 30208
45
+ HIVE: 29839
46
+ IGNITE: 29027
47
+ KAFKA: 19908
48
+ SPARK: 59504
49
+ sub_collections: null
50
+ batches: null
51
+ n_ext: 1
52
+ usable_unit: null
53
+ form:
54
+ by_ext: {txt: 1179062}
55
+ bytes_p50: 341
56
+ bytes_p95: 3408
57
+ bytes_max: 8577459
58
+ size_quantiles:
59
+ max: 8577459
60
+ min: 0
61
+ p25: 138
62
+ p50: 341
63
+ p75: 822
64
+ p90: 1920
65
+ p95: 3408
66
+ p99: 10016
67
+ p99.9: 56454
68
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
69
+ by_ext_size:
70
+ - {ext: txt, rows: 1179062, bytes: 1277864423, p50: 341, p95: 3408, max: 8577459}
71
+ over_cap:
72
+ cap: 61944279
73
+ rows_over: 0
74
+ bytes_over: 0
75
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
76
+ largest:
77
+ - {pointer_id: "apache-jira:NETBEANS:NETBEANS-1357@1546056301", size_bytes: 8577459}
78
+ - {pointer_id: "apache-jira:NETBEANS:NETBEANS-611@1522983216", size_bytes: 4380446}
79
+ - {pointer_id: "apache-jira:NETBEANS:NETBEANS-609@1546055832", size_bytes: 4029103}
80
+ - {pointer_id: "apache-jira:AMQ:AMQ-6705@1497617763", size_bytes: 3051514}
81
+ - {pointer_id: "apache-jira:NETBEANS:NETBEANS-2963@1565775737", size_bytes: 2851879}
82
+ date_years: null
83
+ media_type: {text/plain: 1179062}
84
+ facets:
85
+ issuetype:
86
+ Brainstorming: 249
87
+ Bug: 572215
88
+ Dependency: 201
89
+ "Dependency upgrade": 9145
90
+ Documentation: 2657
91
+ Epic: 1943
92
+ Improvement: 298309
93
+ "New Feature": 69255
94
+ Question: 2512
95
+ Request: 119
96
+ Story: 2665
97
+ Sub-task: 113758
98
+ "Suitable Name Search": 258
99
+ Task: 81116
100
+ "Technical Debt": 1189
101
+ "Technical task": 1419
102
+ Temp: 46
103
+ Test: 12932
104
+ Umbrella: 950
105
+ Wish: 7948
106
+ resolution:
107
+ Abandoned: 6795
108
+ "Auto Closed": 2702
109
+ "Cannot Reproduce": 17624
110
+ Done: 18825
111
+ Duplicate: 48230
112
+ "Feedback Received": 1027
113
+ Fixed: 755444
114
+ Implemented: 9103
115
+ Incomplete: 11141
116
+ "Information Provided": 2388
117
+ Invalid: 22453
118
+ Later: 7297
119
+ "Not A Bug": 3444
120
+ "Not A Problem": 26573
121
+ "Pending Closed": 301
122
+ Resolved: 4028
123
+ "Won't Do": 4088
124
+ "Won't Fix": 47354
125
+ Workaround: 637
126
+ "Works for Me": 334
127
+ status:
128
+ Accepted: 858
129
+ "Awaiting Feedback": 32
130
+ Blocked: 12
131
+ "Changes Suggested": 25
132
+ Closed: 535102
133
+ Continued: 18
134
+ Done: 2930
135
+ "For Review": 18
136
+ "In Progress": 6584
137
+ "In Review": 64
138
+ "Needs Committer": 22
139
+ Open: 165975
140
+ "Patch Available": 6480
141
+ Reopened: 3599
142
+ Resolved: 452535
143
+ "Review In Progress": 42
144
+ Reviewable: 250
145
+ "To Do": 2405
146
+ "Triage Needed": 2039
147
+ WontFix: 40
148
+ artifact: null
149
+ source_origin: null
150
+ parse_cost: null
151
+ parse_quirks: null
152
+ structure:
153
+ self_describing: null
154
+ integrity_builtin: null
155
+ has_reference_output: null
156
+ invariants: null
157
+ tools: null
158
+ ships:
159
+ problem_statement: {present: null}
160
+ reference_solution: {share: null}
161
+ tests: {present: null}
162
+ env_definition: {present: null}
163
+ known_answer: {present: null}
164
+ generator: {present: null}
165
+ upstream_labels: {present: null}
166
+ class: null
167
+ subject:
168
+ domain_span: null
169
+ shape_span: null
170
+ skill_family_span: null
171
+ ont_domain: null
172
+ tech_vocabulary: null
173
+ risk:
174
+ contamination: none
175
+ overlap_check:
176
+ done: true
177
+ pointers_flagged: 1
178
+ pointers_removed: 0
179
+ by_class: {trivial_small: 1}
180
+ benches: {Terminal-Bench: 1}
181
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
182
+ pii: null
183
+ pii_evidence: null
184
+ malware: null
185
+ ship_bytes: unknown
186
+ subset:
187
+ rule: "take all rows; licence, size and version are labels, not filters"
188
+ rows_in_subset: 1179062
189
+ last_built:
190
+ date: "2026-10-02"
191
+ rule: "older rule that filtered by licence tier"
192
+ rows: 1179062
193
+ rows_in_subset: 0
194
+ excluded_by_licence: 1179062
195
+ duplicates: 129461
196
+ access:
197
+ method: "GET uri; the response is JSON."
198
+ verify: "Take fields.description (null counts as an empty string), encode it as UTF-8 and compare size and sha256. The stored file is the description text."
199
+ hosts: {"https://issues.apache.org": 1179062}
200
+ uri_example: "https://issues.apache.org/jira/rest/api/2/issue/AAR-51873?fields=description"
201
+ needs_auth: {"false": 1179062}
202
+ credential: null
203
+ hash_source: {computed: 1179062}
204
+ range_supported: null
205
+ reachable:
206
+ index_flag: {unknown: 1179062}
207
+ index_check:
208
+ checked_at: "2026-10-01"
209
+ sample: 30
210
+ checked: 0
211
+ ok: 0
212
+ by_status: {}
213
+ our_sample: null
214
+ drift: null
215
+ sample: null
216
+ evidence:
217
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ class: {fill: unfilled, basis: "ships not judged"}
221
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ form.by_ext:
223
+ fill: auto
224
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
225
+ checked_at: "2026-10-03"
226
+ form.date_years:
227
+ fill: auto
228
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
229
+ checked_at: "2026-10-03"
230
+ form.over_cap:
231
+ fill: auto
232
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
233
+ checked_at: "2026-10-03"
234
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ form.size_quantiles:
237
+ fill: auto
238
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
239
+ checked_at: "2026-10-03"
240
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
241
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
242
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
243
+ provenance.licence_tiers:
244
+ fill: auto
245
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
246
+ checked_at: "2026-10-03"
247
+ provenance.spdx_top:
248
+ fill: auto
249
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
250
+ checked_at: "2026-10-03"
251
+ risk.contamination:
252
+ fill: auto
253
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
254
+ checked_at: "2026-10-03"
255
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
257
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
259
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
260
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
261
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
262
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
265
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
266
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
268
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
270
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ snapshot.rev:
272
+ fill: auto
273
+ basis: "last commit listed in the publish receipt of this source"
274
+ checked_at: "2026-10-03"
275
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ subset:
286
+ fill: auto
287
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
288
+ checked_at: "2026-10-03"
289
+ unit.pointer_count:
290
+ fill: auto
291
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
292
+ checked_at: "2026-10-03"
293
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
294
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
295
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/arquivo-pt.yaml ADDED
@@ -0,0 +1,354 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: arquivo-pt
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers: {unknown/no: 56284}
29
+ spdx_top: {"(none)": 56284}
30
+ spdx_distinct: 1
31
+ licence_sample_check: null
32
+ unit:
33
+ pointer_level: null
34
+ pointer_count: 56284
35
+ units: {file: 56284}
36
+ state: null
37
+ expanded_count: null
38
+ bytes_total: 2473063538
39
+ rows_without_size: 53899
40
+ collections:
41
+ distinct: 5635
42
+ top:
43
+ www.bportugal.pt: 2943
44
+ www.cgd.pt: 2432
45
+ www.cm-braga.pt: 2336
46
+ www.cm-coimbra.pt: 2399
47
+ www.cm-lisboa.pt: 3335
48
+ www.cp.pt: 2089
49
+ www.ine.pt: 2392
50
+ www.ualg.pt: 2270
51
+ www.uc.pt: 2400
52
+ www.unl.pt: 2313
53
+ sub_collections: null
54
+ batches: null
55
+ n_ext: 11
56
+ usable_unit: null
57
+ form:
58
+ by_ext:
59
+ pdf: 44535
60
+ doc: 4773
61
+ xls: 2489
62
+ ppt: 2216
63
+ ics: 1554
64
+ docx: 280
65
+ rtf: 168
66
+ xlsx: 157
67
+ pptx: 53
68
+ ods: 30
69
+ odt: 29
70
+ bytes_p50: 196608
71
+ bytes_p95: 4187347
72
+ bytes_max: 94315791
73
+ size_quantiles:
74
+ max: 94315791
75
+ min: 1
76
+ p25: 61409
77
+ p50: 196608
78
+ p75: 655258
79
+ p90: 2290769
80
+ p95: 4187347
81
+ p99: 14353456
82
+ p99.9: 52632706
83
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
84
+ by_ext_size:
85
+ - {ext: pdf, rows: 44535, bytes: 2105957769, p50: 242589, p95: 4518852, max: 94315791}
86
+ - {ext: doc, rows: 4773, bytes: 70466788, p50: 58880, p95: 1206272, max: 17609728}
87
+ - {ext: xls, rows: 2489, bytes: 20451840, p50: 56320, p95: 1024000, max: 2698752}
88
+ - {ext: ppt, rows: 2216, bytes: 250276864, p50: 576512, p95: 7534592, max: 65279488}
89
+ - {ext: ics, rows: 1554, bytes: 30713, p50: 397, p95: 1602, max: 3221}
90
+ - {ext: docx, rows: 280, bytes: 430722, p50: 71039, p95: 284534, max: 284534}
91
+ - {ext: rtf, rows: 168, bytes: 6132615, p50: 59904, p95: 5423354, max: 5423354}
92
+ - {ext: xlsx, rows: 157, bytes: 3727797, p50: 123576, p95: 3014798, max: 3014798}
93
+ - {ext: pptx, rows: 53, bytes: 15582184, p50: 2908984, p95: 10077357, max: 10077357}
94
+ - {ext: ods, rows: 30, bytes: 0, p50: null, p95: null, max: null}
95
+ - {ext: odt, rows: 29, bytes: 6246, p50: 6246, p95: 6246, max: 6246}
96
+ over_cap:
97
+ cap: 61944279
98
+ rows_over: 2
99
+ bytes_over: 159595279
100
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
101
+ largest:
102
+ - pointer_id: "arquivo-pt:www.cm-braga.pt:20251019230046/d5fa56c8dd-1.1-planta_classificacao_qualificacao_solo.pdf"
103
+ size_bytes: 94315791
104
+ - pointer_id: "arquivo-pt:www.cm-lisboa.pt:20100809112532/998866e7a6-Concursos_reuniao_camara_24.3.2010.ppt"
105
+ size_bytes: 65279488
106
+ - pointer_id: "arquivo-pt:www.cm-braga.pt:20240521011817/1c815c4ce0-Relatorio_da_Reviso_do_PDM.pdf"
107
+ size_bytes: 52632706
108
+ - pointer_id: "arquivo-pt:www.cm-lisboa.pt:20090929025710/8d8ef96c68-Abertura_Escola_Armador.ppt"
109
+ size_bytes: 46785024
110
+ - pointer_id: "arquivo-pt:www.cgd.pt:20181102080222/ec6ea078d4-CGD-Sustainability-Report-2011.pdf"
111
+ size_bytes: 38416680
112
+ date_years:
113
+ key: origin.captured_at
114
+ rows_with_date: 56284
115
+ counts:
116
+ "2000": 3
117
+ "2001": 127
118
+ "2002": 91
119
+ "2003": 1076
120
+ "2004": 1617
121
+ "2005": 1697
122
+ "2006": 1010
123
+ "2007": 2327
124
+ "2008": 9320
125
+ "2009": 2901
126
+ "2010": 3207
127
+ "2011": 3508
128
+ "2012": 1904
129
+ "2013": 2163
130
+ "2014": 1401
131
+ "2015": 2323
132
+ "2016": 3690
133
+ "2017": 2625
134
+ "2018": 2644
135
+ "2019": 2635
136
+ "2020": 3344
137
+ "2021": 1367
138
+ "2022": 442
139
+ "2023": 774
140
+ "2024": 1424
141
+ "2025": 2481
142
+ "2026": 183
143
+ media_type:
144
+ application/msword: 4773
145
+ application/pdf: 44535
146
+ application/rtf: 168
147
+ application/vnd.ms-excel: 2489
148
+ application/vnd.ms-powerpoint: 2216
149
+ application/vnd.oasis.opendocument.spreadsheet: 30
150
+ application/vnd.openxmlformats-officedocument.presentationml.presentation: 53
151
+ application/vnd.openxmlformats-officedocument.spreadsheetml.sheet: 157
152
+ application/vnd.openxmlformats-officedocument.wordprocessingml.document: 280
153
+ text/calendar: 1554
154
+ facets:
155
+ head_content_type:
156
+ application/DOC: 3
157
+ application/doc: 1
158
+ application/msword: 231
159
+ application/pdf: 2174
160
+ "application/pdf; Charset=UTF-8": 1
161
+ "application/pdf; charset=utf-8": 48
162
+ application/ppt: 1
163
+ application/rtf: 6
164
+ application/vnd.ms-excel: 130
165
+ application/vnd.ms-powerpoint: 96
166
+ application/vnd.oasis.opendocument.text: 1
167
+ application/xls: 7
168
+ "application/xls; charset=utf-8": 1
169
+ text/calendar: 4
170
+ "text/calendar; charset=UTF-8": 26
171
+ "text/calendar; charset=iso-8859-15": 3
172
+ "text/calendar; charset=utf-8": 51
173
+ "text/calendar;charset=UTF-8": 1
174
+ text/ics: 2
175
+ text/rtf: 3
176
+ head_status: {"200": 2809, "302": 1, net_error: 1}
177
+ mime:
178
+ application/doc: 47
179
+ application/docx: 7
180
+ application/msword: 4726
181
+ application/pdf: 44535
182
+ application/ppt: 39
183
+ application/rtf: 117
184
+ application/vnd.ms-excel: 2401
185
+ application/vnd.ms-powerpoint: 2177
186
+ application/vnd.oasis.opendocument.text: 29
187
+ application/xls: 88
188
+ application/xlsx: 1
189
+ text/calendar: 1518
190
+ text/ics: 36
191
+ text/rtf: 51
192
+ office_type:
193
+ doc: 4773
194
+ docx: 280
195
+ ics: 1554
196
+ ods: 30
197
+ odt: 29
198
+ pdf: 44535
199
+ ppt: 2216
200
+ pptx: 53
201
+ rtf: 168
202
+ xls: 2489
203
+ xlsx: 157
204
+ seed_domain:
205
+ anmp.pt: 1469
206
+ bportugal.pt: 2945
207
+ cgd.pt: 2431
208
+ cm-braga.pt: 2336
209
+ cm-coimbra.pt: 2434
210
+ cm-lisboa.pt: 3325
211
+ cp.pt: 2186
212
+ ctt.pt: 1594
213
+ edp.pt: 1507
214
+ gov.pt: 2066
215
+ ine.pt: 2391
216
+ ipp.pt: 1885
217
+ iscte-iul.pt: 1904
218
+ presidencia.pt: 1293
219
+ ualg.pt: 2266
220
+ uc.pt: 2395
221
+ uevora.pt: 1609
222
+ ulisboa.pt: 1797
223
+ uminho.pt: 1972
224
+ unl.pt: 2327
225
+ size_source: {http_head: 2385, none: 53899}
226
+ via: {cdx: 46946, textsearch: 9338}
227
+ artifact: null
228
+ source_origin: null
229
+ parse_cost: null
230
+ parse_quirks: null
231
+ structure:
232
+ self_describing: null
233
+ integrity_builtin: null
234
+ has_reference_output: null
235
+ invariants: null
236
+ tools: null
237
+ ships:
238
+ problem_statement: {present: null}
239
+ reference_solution: {share: null}
240
+ tests: {present: null}
241
+ env_definition: {present: null}
242
+ known_answer: {present: null}
243
+ generator: {present: null}
244
+ upstream_labels: {present: null}
245
+ class: null
246
+ subject:
247
+ domain_span: null
248
+ shape_span: null
249
+ skill_family_span: null
250
+ ont_domain: null
251
+ tech_vocabulary: null
252
+ risk:
253
+ contamination: null
254
+ overlap_check: null
255
+ pii: null
256
+ pii_evidence: null
257
+ malware: null
258
+ ship_bytes: unknown
259
+ subset:
260
+ rule: "take all rows; licence, size and version are labels, not filters"
261
+ rows_in_subset: 56284
262
+ last_built: null
263
+ access:
264
+ method: null
265
+ verify: null
266
+ hosts: {"https://arquivo.pt": 56284}
267
+ uri_example: "https://arquivo.pt/wayback/20200315141350id_/https://12e03af7-3e7d-c65e-c3b5-d01ce01e03b3.filesusr.com/ugd/965fea_1bee472c1f434cfc827c5659fed8e8c9.doc?dn=NCSNDOCENTES08032019%20(1).doc"
268
+ needs_auth: {"false": 56284}
269
+ credential: null
270
+ hash_source: {computed_on_fetch: 143, source_published: 56141}
271
+ range_supported: null
272
+ reachable:
273
+ index_flag: {"(none)": 56284}
274
+ index_check: null
275
+ our_sample: null
276
+ drift: null
277
+ sample: null
278
+ evidence:
279
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ class: {fill: unfilled, basis: "ships not judged"}
283
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ form.by_ext:
285
+ fill: auto
286
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
287
+ checked_at: "2026-10-03"
288
+ form.date_years:
289
+ fill: auto
290
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
291
+ checked_at: "2026-10-03"
292
+ form.over_cap:
293
+ fill: auto
294
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
295
+ checked_at: "2026-10-03"
296
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
297
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
298
+ form.size_quantiles:
299
+ fill: auto
300
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
301
+ checked_at: "2026-10-03"
302
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
303
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
304
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
305
+ provenance.licence_tiers:
306
+ fill: auto
307
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
308
+ checked_at: "2026-10-03"
309
+ provenance.spdx_top:
310
+ fill: auto
311
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
312
+ checked_at: "2026-10-03"
313
+ risk.contamination: {fill: unfilled, basis: "this source was not part of the benchmark overlap scan"}
314
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
315
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
316
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
317
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
318
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
319
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
320
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
321
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
322
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
323
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
324
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
325
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
326
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
327
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
328
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
329
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
330
+ snapshot.rev:
331
+ fill: auto
332
+ basis: "last commit listed in the publish receipt of this source"
333
+ checked_at: "2026-10-03"
334
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
335
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
336
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
337
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
338
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
339
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
340
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
341
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
342
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
343
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
344
+ subset:
345
+ fill: auto
346
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
347
+ checked_at: "2026-10-03"
348
+ unit.pointer_count:
349
+ fill: auto
350
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
351
+ checked_at: "2026-10-03"
352
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
353
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
354
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/artifact-hub-helm-charts.yaml ADDED
@@ -0,0 +1,259 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: artifact-hub-helm-charts
8
+ parent: null
9
+ title: "Artifact Hub Helm charts"
10
+ description: "The packaged chart (.tgz) of every version of every Helm chart in the Helm repositories listed on Artifact Hub (artifacthub.io), except repositories and download addresses hosted on GitHub and repositories served over oci://. The repository list comes from the Artifact Hub API; versions, download addresses and digests come from each repository's index.yaml. Packages can contain malicious code; treat files as data and run nothing outside a sandbox."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-02"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per chart version (artifacthub.io/license annotation, or the Artifact Hub package license)"
20
+ licence_url: "https://artifacthub.io"
21
+ licence_statement: "Each chart publisher releases its charts under its own license; neither Artifact Hub nor the Helm repository format gives one license for all charts. The license files inside the chart packages were not checked one by one."
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: "license_override.notes gives the artifacthub.io/license annotation of the version, or, where the version has none, the license Artifact Hub shows for the chart (that of its latest version; source_meta.license_source says which); license_override.spdx gives the SPDX expression where one is recognised."
27
+ licence_tiers: {unknown/unknown: 104262, yes/conditional: 2709, yes/yes: 20310}
28
+ spdx_top:
29
+ "(none)": 104262
30
+ AGPL-3.0: 136
31
+ AGPL-3.0-only: 74
32
+ Apache-2.0: 15444
33
+ BSD-2-Clause-Patent: 7
34
+ BSD-3-Clause: 7
35
+ EPL-2.0: 1
36
+ GPL-2.0: 30
37
+ GPL-2.0-or-later: 3
38
+ GPL-3.0: 351
39
+ GPL-3.0-only: 36
40
+ GPL-3.0-or-later: 1212
41
+ LGPL-3.0: 9
42
+ LGPL-3.0-only: 374
43
+ MIT: 4852
44
+ spdx_distinct: 17
45
+ licence_sample_check: null
46
+ unit:
47
+ pointer_level: null
48
+ pointer_count: 127281
49
+ units: {file: 127281}
50
+ state: null
51
+ expanded_count: null
52
+ bytes_total: 0
53
+ rows_without_size: 127281
54
+ collections:
55
+ distinct: 456
56
+ top:
57
+ appscode: 4918
58
+ bitnami: 14084
59
+ crossplane: 3379
60
+ graviteeio: 2373
61
+ jfrog: 4372
62
+ loft: 4027
63
+ nest-chart: 11652
64
+ supporttools: 2144
65
+ teleport-agent-kube: 5082
66
+ wenerme: 8650
67
+ sub_collections: null
68
+ batches: null
69
+ n_ext: 1
70
+ usable_unit: null
71
+ form:
72
+ by_ext: {tgz: 127281}
73
+ bytes_p50: null
74
+ bytes_p95: null
75
+ bytes_max: null
76
+ size_quantiles: null
77
+ by_ext_size:
78
+ - {ext: tgz, rows: 127281, bytes: 0, p50: null, p95: null, max: null}
79
+ over_cap:
80
+ cap: 61944279
81
+ rows_over: 0
82
+ bytes_over: 0
83
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
84
+ largest: []
85
+ date_years: null
86
+ media_type: {"(none)": 127281}
87
+ facets:
88
+ deprecated: {"false": 126581, "true": 700}
89
+ license:
90
+ AGPL-3.0: 136
91
+ AGPL-3.0-only: 74
92
+ Apache-2.0: 15444
93
+ BSD-2-Clause-Patent: 7
94
+ BSD-3-Clause: 7
95
+ BUSL-1.1: 24
96
+ Elastic-2.0: 306
97
+ GPL-2.0: 30
98
+ GPL-3.0: 351
99
+ GPL-3.0-only: 36
100
+ GPL-3.0-or-later: 1212
101
+ LGPL-3.0: 9
102
+ LGPL-3.0-only: 374
103
+ MIT: 4791
104
+ "MIT AND BUSL-1.1": 19
105
+ "MIT License": 61
106
+ MIT-open-group: 61
107
+ MPL-2.0: 344
108
+ MPL-2.0-no-copyleft-exception: 139
109
+ deprecated_GPL-3.0+: 361
110
+ license_source: {artifacthub_package: 14860, chart_annotation: 8933}
111
+ official: {"false": 121296, "true": 5985}
112
+ verified_publisher: {"false": 69086, "true": 58195}
113
+ artifact: null
114
+ source_origin: null
115
+ parse_cost: null
116
+ parse_quirks: null
117
+ structure:
118
+ self_describing: null
119
+ integrity_builtin: null
120
+ has_reference_output: null
121
+ invariants: null
122
+ tools: null
123
+ ships:
124
+ problem_statement: {present: null}
125
+ reference_solution: {share: null}
126
+ tests: {present: null}
127
+ env_definition: {present: null}
128
+ known_answer: {present: null}
129
+ generator: {present: null}
130
+ upstream_labels: {present: null}
131
+ class: null
132
+ subject:
133
+ domain_span: null
134
+ shape_span: null
135
+ skill_family_span: null
136
+ ont_domain: null
137
+ tech_vocabulary: null
138
+ risk:
139
+ contamination: none
140
+ overlap_check:
141
+ done: true
142
+ pointers_flagged: 0
143
+ pointers_removed: 0
144
+ by_class: {}
145
+ benches: {}
146
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
147
+ pii: null
148
+ pii_evidence: null
149
+ malware: null
150
+ ship_bytes: unknown
151
+ subset:
152
+ rule: "take all rows; licence, size and version are labels, not filters"
153
+ rows_in_subset: 127281
154
+ last_built: null
155
+ access:
156
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
157
+ verify: "Compare sha256 where given (published as digest in index.yaml; hash_source source_published). Versions without a digest have no published hash (computed_on_fetch). The size is not published (size_bytes null); record the measured size."
158
+ hosts:
159
+ "https://charts.appscode.com": 4918
160
+ "https://charts.bitnami.com": 14085
161
+ "https://charts.crossplane.io": 3540
162
+ "https://charts.helm.sh": 12884
163
+ "https://charts.jfrog.io": 4372
164
+ "https://charts.loft.sh": 4027
165
+ "https://charts.releases.teleport.dev": 5082
166
+ "https://charts.wener.tech": 8650
167
+ "https://gitlab-charts.s3.amazonaws.com": 3018
168
+ "https://helm.gravitee.io": 2373
169
+ uri_example: "https://gitlab.trofimoff.net/api/v4/projects/32/packages/helm/stable/charts/adguardhome-0.1.10.tgz"
170
+ needs_auth: {"false": 127281}
171
+ credential: null
172
+ hash_source: {computed_on_fetch: 2123, source_published: 125158}
173
+ range_supported: null
174
+ reachable:
175
+ index_flag: {unknown: 127281}
176
+ index_check: null
177
+ our_sample: null
178
+ drift: null
179
+ sample: null
180
+ evidence:
181
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
184
+ class: {fill: unfilled, basis: "ships not judged"}
185
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
186
+ form.by_ext:
187
+ fill: auto
188
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
189
+ checked_at: "2026-10-03"
190
+ form.date_years:
191
+ fill: auto
192
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
193
+ checked_at: "2026-10-03"
194
+ form.over_cap:
195
+ fill: auto
196
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
197
+ checked_at: "2026-10-03"
198
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
199
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ form.size_quantiles:
201
+ fill: auto
202
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
203
+ checked_at: "2026-10-03"
204
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
206
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ provenance.licence_tiers:
208
+ fill: auto
209
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
210
+ checked_at: "2026-10-03"
211
+ provenance.spdx_top:
212
+ fill: auto
213
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
214
+ checked_at: "2026-10-03"
215
+ risk.contamination:
216
+ fill: auto
217
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
218
+ checked_at: "2026-10-03"
219
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
223
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
228
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ snapshot.rev:
236
+ fill: auto
237
+ basis: "last commit listed in the publish receipt of this source"
238
+ checked_at: "2026-10-03"
239
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
240
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
241
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
242
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
243
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
244
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
245
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
247
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
249
+ subset:
250
+ fill: auto
251
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
252
+ checked_at: "2026-10-03"
253
+ unit.pointer_count:
254
+ fill: auto
255
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
256
+ checked_at: "2026-10-03"
257
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
259
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/bioinformatics-public-archives.yaml ADDED
@@ -0,0 +1,305 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: bioinformatics-public-archives
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers: {yes/conditional: 20, yes/yes: 239778}
29
+ spdx_top: {"(none)": 2233, CC-BY-SA-3.0: 20, CC0-1.0: 237545}
30
+ spdx_distinct: 3
31
+ licence_sample_check: null
32
+ unit:
33
+ pointer_level: null
34
+ pointer_count: 239798
35
+ units: {file: 239798}
36
+ state: null
37
+ expanded_count: null
38
+ bytes_total: 2796870166887
39
+ rows_without_size: 237457
40
+ collections:
41
+ distinct: 5
42
+ top: {chembl: 20, ensembl: 2233, interpro: 72, pdb: 237457, pfam: 16}
43
+ sub_collections: null
44
+ batches: null
45
+ n_ext: 15
46
+ usable_unit: null
47
+ form:
48
+ by_ext:
49
+ gz: 239473
50
+ "(none)": 180
51
+ parquet: 57
52
+ fai: 35
53
+ gzi: 35
54
+ txt: 7
55
+ dat: 2
56
+ dtd: 2
57
+ attribution: 1
58
+ h5: 1
59
+ html: 1
60
+ list: 1
61
+ pdf: 1
62
+ tgz: 1
63
+ xml: 1
64
+ bytes_p50: 12907160
65
+ bytes_p95: 472829272
66
+ bytes_max: 2066193325033
67
+ size_quantiles:
68
+ max: 2066193325033
69
+ min: 114
70
+ p25: 6236007
71
+ p50: 12907160
72
+ p75: 27684504
73
+ p90: 271090428
74
+ p95: 472829272
75
+ p99: 870202217
76
+ p99.9: 57135455149
77
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
78
+ by_ext_size:
79
+ - {ext: gz, rows: 239473, bytes: 2766852836458, p50: 14711160, p95: 378358957, max: 2066193325033}
80
+ - {ext: "(none)", rows: 180, bytes: 3561120, p50: 2432, p95: 4985, max: 3064685}
81
+ - {ext: parquet, rows: 57, bytes: 29593081698, p50: 549475950, p95: 676464572, max: 744688702}
82
+ - {ext: fai, rows: 35, bytes: 29294466, p50: 92057, p95: 2819431, max: 16161318}
83
+ - {ext: gzi, rows: 35, bytes: 14719176, p50: 306984, p95: 783160, max: 795048}
84
+ - {ext: txt, rows: 7, bytes: 2080307, p50: 25592, p95: 1254330, max: 1254330}
85
+ - {ext: dat, rows: 2, bytes: 3808633, p50: 1266970, p95: 2541663, max: 2541663}
86
+ - {ext: dtd, rows: 2, bytes: 6085, p50: 898, p95: 5187, max: 5187}
87
+ - {ext: attribution, rows: 1, bytes: 733, p50: 733, p95: 733, max: 733}
88
+ - {ext: h5, rows: 1, bytes: 329575511, p50: 329575511, p95: 329575511, max: 329575511}
89
+ - {ext: html, rows: 1, bytes: 117000, p50: 117000, p95: 117000, max: 117000}
90
+ - {ext: list, rows: 1, bytes: 2989742, p50: 2989742, p95: 2989742, max: 2989742}
91
+ - {ext: pdf, rows: 1, bytes: 819544, p50: 819544, p95: 819544, max: 819544}
92
+ - {ext: tgz, rows: 1, bytes: 33366220, p50: 33366220, p95: 33366220, max: 33366220}
93
+ - {ext: xml, rows: 1, bytes: 3910194, p50: 3910194, p95: 3910194, max: 3910194}
94
+ over_cap:
95
+ cap: 61944279
96
+ rows_over: 288
97
+ bytes_over: 2767544466809
98
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
99
+ largest:
100
+ - pointer_id: "bioinformatics-public-archives:interpro:matches-api-data.tar.gz"
101
+ size_bytes: 2066193325033
102
+ - pointer_id: "bioinformatics-public-archives:interpro:uniparc_match.tar.gz"
103
+ size_bytes: 419004150100
104
+ - pointer_id: "bioinformatics-public-archives:interpro:match_complete.xml.gz"
105
+ size_bytes: 57135455149
106
+ - {pointer_id: "bioinformatics-public-archives:interpro:interpro-n.xml.gz", size_bytes: 27967782828}
107
+ - {pointer_id: "bioinformatics-public-archives:pfam:Pfam-A.full.gz", size_bytes: 23991810911}
108
+ date_years: null
109
+ media_type: {"(none)": 239798}
110
+ facets:
111
+ checksum_source: {CHECKSUMS: 2232, checksums.txt: 9, md5_checksums: 14, sidecar_md5: 6}
112
+ format:
113
+ attribution: 1
114
+ dat: 2
115
+ dtd: 2
116
+ fai: 35
117
+ fasta: 1981
118
+ fps: 1
119
+ gz: 53524
120
+ gzi: 35
121
+ hmm: 1
122
+ list: 1
123
+ mmcif: 67056
124
+ none: 180
125
+ parquet: 57
126
+ pdb: 23015
127
+ pdf: 31273
128
+ tar: 70
129
+ text: 9
130
+ tgz: 1
131
+ tsv: 2
132
+ xml: 62549
133
+ kind:
134
+ assembly_mmcif: 20214
135
+ assembly_pdb: 19617
136
+ cdna: 101
137
+ cds: 70
138
+ combined_nmr_data_nef: 29
139
+ dna: 1778
140
+ dna_index: 105
141
+ mmcif: 15636
142
+ ncrna: 70
143
+ nmr_chemical_shifts: 43
144
+ nmr_restraints_v1: 1287
145
+ nmr_restraints_v2: 1286
146
+ pdb: 15155
147
+ pdb_bundle: 55
148
+ pdbml: 15636
149
+ pdbml_extatom: 15636
150
+ pdbml_noatom: 15636
151
+ pep: 98
152
+ structure_factors: 7860
153
+ validation_report: 109338
154
+ site: {chembl: 20, ensembl: 2233, interpro: 72, pdb: 237457, pfam: 16}
155
+ size_source: {head: 2337, listing: 4}
156
+ species:
157
+ acanthochromis_polyacanthus: 20
158
+ ailuropoda_melanoleuca: 83
159
+ amphiprion_ocellaris: 93
160
+ amphiprion_percula: 92
161
+ anabas_testudineus: 87
162
+ anas_platyrhynchos: 143
163
+ anas_platyrhynchos_platyrhynchos: 107
164
+ anolis_carolinensis: 59
165
+ aquila_chrysaetos_chrysaetos: 101
166
+ astatotilapia_calliptera: 84
167
+ astyanax_mexicanus: 95
168
+ balaenoptera_musculus: 89
169
+ betta_splendens: 83
170
+ bos_grunniens: 113
171
+ bos_indicus_hybrid: 110
172
+ bos_taurus: 114
173
+ bos_taurus_gca963921495v1: 114
174
+ bos_taurus_hybrid: 110
175
+ bos_taurus_tuli: 111
176
+ bos_taurus_wagyu: 111
177
+ version: {"110.0": 72, "116": 2233, "2026-09-26T10:14:31Z": 237457, "37": 20, "38.2": 16}
178
+ artifact: null
179
+ source_origin: null
180
+ parse_cost: null
181
+ parse_quirks: null
182
+ structure:
183
+ self_describing: null
184
+ integrity_builtin: null
185
+ has_reference_output: null
186
+ invariants: null
187
+ tools: null
188
+ ships:
189
+ problem_statement: {present: null}
190
+ reference_solution: {share: null}
191
+ tests: {present: null}
192
+ env_definition: {present: null}
193
+ known_answer: {present: null}
194
+ generator: {present: null}
195
+ upstream_labels: {present: null}
196
+ class: null
197
+ subject:
198
+ domain_span: null
199
+ shape_span: null
200
+ skill_family_span: null
201
+ ont_domain: null
202
+ tech_vocabulary: null
203
+ risk:
204
+ contamination: null
205
+ overlap_check: null
206
+ pii: null
207
+ pii_evidence: null
208
+ malware: null
209
+ ship_bytes: unknown
210
+ subset:
211
+ rule: "take all rows; licence, size and version are labels, not filters"
212
+ rows_in_subset: 239798
213
+ last_built: null
214
+ access:
215
+ method: null
216
+ verify: null
217
+ hosts: {"https://files.wwpdb.org": 237457, "https://ftp.ebi.ac.uk": 108, "https://ftp.ensembl.org": 2233}
218
+ uri_example: "https://ftp.ebi.ac.uk/pub/databases/chembl/ChEMBLdb/releases/chembl_37/LICENSE"
219
+ needs_auth: {"false": 239798}
220
+ credential: null
221
+ hash_source: {computed_on_fetch: 239769, source_published: 29}
222
+ range_supported: null
223
+ reachable:
224
+ index_flag: {unknown: 239798}
225
+ index_check: null
226
+ our_sample: null
227
+ drift: null
228
+ sample: null
229
+ evidence:
230
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ class: {fill: unfilled, basis: "ships not judged"}
234
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ form.by_ext:
236
+ fill: auto
237
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
238
+ checked_at: "2026-10-03"
239
+ form.date_years:
240
+ fill: auto
241
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
242
+ checked_at: "2026-10-03"
243
+ form.over_cap:
244
+ fill: auto
245
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
246
+ checked_at: "2026-10-03"
247
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
249
+ form.size_quantiles:
250
+ fill: auto
251
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
252
+ checked_at: "2026-10-03"
253
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
255
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ provenance.licence_tiers:
257
+ fill: auto
258
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
259
+ checked_at: "2026-10-03"
260
+ provenance.spdx_top:
261
+ fill: auto
262
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
263
+ checked_at: "2026-10-03"
264
+ risk.contamination: {fill: unfilled, basis: "this source was not part of the benchmark overlap scan"}
265
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
266
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
268
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
269
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
270
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
272
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
273
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
275
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ snapshot.rev:
282
+ fill: auto
283
+ basis: "last commit listed in the publish receipt of this source"
284
+ checked_at: "2026-10-03"
285
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
287
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
290
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
293
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
294
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
295
+ subset:
296
+ fill: auto
297
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
298
+ checked_at: "2026-10-03"
299
+ unit.pointer_count:
300
+ fill: auto
301
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
302
+ checked_at: "2026-10-03"
303
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
304
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
305
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/centos-cbs-build-logs.yaml ADDED
@@ -0,0 +1,241 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: centos-cbs-build-logs
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers: {unknown/unknown: 61178}
29
+ spdx_top: {"(none)": 61178}
30
+ spdx_distinct: 1
31
+ licence_sample_check: null
32
+ unit:
33
+ pointer_level: null
34
+ pointer_count: 61178
35
+ units: {file: 61178}
36
+ state: null
37
+ expanded_count: null
38
+ bytes_total: 106318040252
39
+ rows_without_size: 0
40
+ collections:
41
+ distinct: 2
42
+ top: {complete: 60907, failed: 271}
43
+ sub_collections: null
44
+ batches: null
45
+ n_ext: 1
46
+ usable_unit: null
47
+ form:
48
+ by_ext: {log: 61178}
49
+ bytes_p50: 52376
50
+ bytes_p95: 7004464
51
+ bytes_max: 330491223
52
+ size_quantiles:
53
+ max: 330491223
54
+ min: 72
55
+ p25: 18648
56
+ p50: 52376
57
+ p75: 2494223
58
+ p90: 6205753
59
+ p95: 7004464
60
+ p99: 9666461
61
+ p99.9: 46464012
62
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
63
+ by_ext_size:
64
+ - {ext: log, rows: 61178, bytes: 106318040252, p50: 52376, p95: 7004464, max: 330491223}
65
+ over_cap:
66
+ cap: 61944279
67
+ rows_over: 44
68
+ bytes_over: 4989857669
69
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
70
+ largest:
71
+ - {pointer_id: "centos-cbs-build-logs:complete:44824/x86_64/build.log", size_bytes: 330491223}
72
+ - {pointer_id: "centos-cbs-build-logs:complete:47378/x86_64/build.log", size_bytes: 328735539}
73
+ - {pointer_id: "centos-cbs-build-logs:complete:46499/x86_64/build.log", size_bytes: 269532991}
74
+ - {pointer_id: "centos-cbs-build-logs:complete:48978/x86_64/build.log", size_bytes: 268080115}
75
+ - {pointer_id: "centos-cbs-build-logs:complete:75683/riscv64/build.log", size_bytes: 208588546}
76
+ date_years:
77
+ key: start_time
78
+ rows_with_date: 61178
79
+ counts: {"2023": 12932, "2024": 12066, "2025": 15136, "2026": 21044}
80
+ media_type: {text/plain: 61178}
81
+ facets:
82
+ arch:
83
+ aarch64: 15616
84
+ image/aarch64: 167
85
+ image/riscv64: 4
86
+ image/x86_64: 170
87
+ noarch: 10062
88
+ ppc64le: 9590
89
+ riscv64: 5405
90
+ s390x: 515
91
+ x86_64: 19649
92
+ build_state: {COMPLETE: 60907, FAILED: 271}
93
+ log_name:
94
+ build.log: 60935
95
+ dnf.librepo.log: 28
96
+ dnf.log: 28
97
+ dnf.rpm.log: 28
98
+ hw_info.log: 28
99
+ installed_pkgs.log: 19
100
+ mock_config.log: 28
101
+ mock_output.log: 28
102
+ root.log: 28
103
+ state.log: 28
104
+ task_state: {CANCELED: 20, FAILED: 251}
105
+ artifact: null
106
+ source_origin: null
107
+ parse_cost: null
108
+ parse_quirks: null
109
+ structure:
110
+ self_describing: null
111
+ integrity_builtin: null
112
+ has_reference_output: null
113
+ invariants: null
114
+ tools: null
115
+ ships:
116
+ problem_statement: {present: null}
117
+ reference_solution: {share: null}
118
+ tests: {present: null}
119
+ env_definition: {present: null}
120
+ known_answer: {present: null}
121
+ generator: {present: null}
122
+ upstream_labels: {present: null}
123
+ class: null
124
+ subject:
125
+ domain_span: null
126
+ shape_span: null
127
+ skill_family_span: null
128
+ ont_domain: null
129
+ tech_vocabulary: null
130
+ risk:
131
+ contamination: none
132
+ overlap_check:
133
+ done: true
134
+ pointers_flagged: 0
135
+ pointers_removed: 0
136
+ by_class: {}
137
+ benches: {}
138
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
139
+ pii: null
140
+ pii_evidence: null
141
+ malware: null
142
+ ship_bytes: unknown
143
+ subset:
144
+ rule: "take all rows; licence, size and version are labels, not filters"
145
+ rows_in_subset: 61178
146
+ last_built: null
147
+ access:
148
+ method: null
149
+ verify: null
150
+ hosts: {"https://cbs.centos.org": 61178}
151
+ uri_example: "https://cbs.centos.org/kojifiles/packages/libsodium/1.0.18/8.el9s/data/logs/aarch64/build.log"
152
+ needs_auth: {"false": 61178}
153
+ credential: null
154
+ hash_source: {computed_on_fetch: 61178}
155
+ range_supported: null
156
+ reachable:
157
+ index_flag: {unknown: 61178}
158
+ index_check: null
159
+ our_sample: null
160
+ drift: null
161
+ sample: null
162
+ evidence:
163
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
164
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
165
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
166
+ class: {fill: unfilled, basis: "ships not judged"}
167
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
168
+ form.by_ext:
169
+ fill: auto
170
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
171
+ checked_at: "2026-10-03"
172
+ form.date_years:
173
+ fill: auto
174
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
175
+ checked_at: "2026-10-03"
176
+ form.over_cap:
177
+ fill: auto
178
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
179
+ checked_at: "2026-10-03"
180
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
181
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ form.size_quantiles:
183
+ fill: auto
184
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
185
+ checked_at: "2026-10-03"
186
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
188
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
189
+ provenance.licence_tiers:
190
+ fill: auto
191
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
192
+ checked_at: "2026-10-03"
193
+ provenance.spdx_top:
194
+ fill: auto
195
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
196
+ checked_at: "2026-10-03"
197
+ risk.contamination:
198
+ fill: auto
199
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
200
+ checked_at: "2026-10-03"
201
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
205
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
212
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
213
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
214
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ snapshot.rev:
218
+ fill: auto
219
+ basis: "last commit listed in the publish receipt of this source"
220
+ checked_at: "2026-10-03"
221
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
228
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ subset:
232
+ fill: auto
233
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
234
+ checked_at: "2026-10-03"
235
+ unit.pointer_count:
236
+ fill: auto
237
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
238
+ checked_at: "2026-10-03"
239
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
240
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
241
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/chromium-luci-builds.yaml ADDED
@@ -0,0 +1,244 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: chromium-luci-builds
8
+ parent: null
9
+ title: "Chromium LUCI failed build logs"
10
+ description: "Failed builds (status FAILURE or INFRA_FAILURE) of the Chromium project on LUCI (all buckets) created before 2026-10-01T23:20:52Z, listed through the Buildbucket pRPC API (Builds.SearchBuilds, per builder, newest first). One pointer per failed step: the step's stdout log as raw text from LogDog. Builder, build id and number, failed step, failure summary, times and the triggering commit or Gerrit change are in source_meta."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "No license stated"
20
+ licence_url: "https://cr-buildbucket.appspot.com/rpcexplorer/"
21
+ licence_statement: "The logs are machine output of LUCI builds. The operator states no license for them; the Chromium source code license (BSD-3-Clause) does not cover build logs."
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {unknown/unknown: 8665027}
28
+ spdx_top: {"(none)": 8665027}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 8665027
34
+ units: {file: 8665027}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 0
38
+ rows_without_size: 8665027
39
+ collections:
40
+ distinct: 9
41
+ top:
42
+ build: 50975
43
+ ci: 2811482
44
+ codesearch: 56
45
+ infra: 2
46
+ reviver: 68160
47
+ ssci: 2
48
+ try: 5729604
49
+ webrtc: 2617
50
+ webrtc.fyi: 2129
51
+ sub_collections: null
52
+ batches: null
53
+ n_ext: 1
54
+ usable_unit: null
55
+ form:
56
+ by_ext: {log: 8665027}
57
+ bytes_p50: null
58
+ bytes_p95: null
59
+ bytes_max: null
60
+ size_quantiles: null
61
+ by_ext_size:
62
+ - {ext: log, rows: 8665027, bytes: 0, p50: null, p95: null, max: null}
63
+ over_cap:
64
+ cap: 61944279
65
+ rows_over: 0
66
+ bytes_over: 0
67
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
68
+ largest: []
69
+ date_years:
70
+ key: start_time
71
+ rows_with_date: 8665027
72
+ counts: {"2025": 3667714, "2026": 4997313}
73
+ media_type: {text/plain: 8665027}
74
+ facets:
75
+ bucket:
76
+ build: 50975
77
+ ci: 2811482
78
+ codesearch: 56
79
+ infra: 2
80
+ reviver: 68160
81
+ ssci: 2
82
+ try: 5729604
83
+ webrtc: 2617
84
+ webrtc.fyi: 2129
85
+ status: {FAILURE: 8028722, INFRA_FAILURE: 636305}
86
+ step_status: {FAILURE: 8046841, INFRA_FAILURE: 618186}
87
+ summary_markdown:
88
+ "1 total failed UTR runs": 80
89
+ "2 total failed UTR runs": 184
90
+ "Error: Invalid targets": 218
91
+ "HTTP status (404)": 204
92
+ "HTTP status (404)\nHTTP status (404)": 97
93
+ "LUCI CV no longer needs this Tryjob": 207
94
+ "Step('Generate diffs') (retcode: 1)": 41
95
+ "Step('analyze') (retcode: 1)": 134
96
+ "Step('create bundle') (retcode: 1)": 47
97
+ "Step('download corpora') (retcode: 1)": 678
98
+ "Step('fetch ANGLE DEPS') (retcode: 1)": 215
99
+ "Step('fetch Chromium DEPS') (retcode: 1)": 103
100
+ "Step('fetch rejections') (retcode: 1)": 52
101
+ "Step('gclient runhooks') (retcode: 2)": 111607
102
+ "Step('isolate tests') (retcode: 1)": 725
103
+ "Step('metrics.metrics') (retcode: 2)": 240
104
+ "Step('package clang') (retcode: 1)": 2540
105
+ "Step('package rust') (retcode: 1)": 3776
106
+ "Step('run coverage script') (retcode: 1)": 65587
107
+ "Step('run eval prompts') (retcode: 1)": 1601
108
+ artifact: null
109
+ source_origin: null
110
+ parse_cost: null
111
+ parse_quirks: null
112
+ structure:
113
+ self_describing: null
114
+ integrity_builtin: null
115
+ has_reference_output: null
116
+ invariants: null
117
+ tools: null
118
+ ships:
119
+ problem_statement: {present: null}
120
+ reference_solution: {share: null}
121
+ tests: {present: null}
122
+ env_definition: {present: null}
123
+ known_answer: {present: null}
124
+ generator: {present: null}
125
+ upstream_labels: {present: null}
126
+ class: null
127
+ subject:
128
+ domain_span: null
129
+ shape_span: null
130
+ skill_family_span: null
131
+ ont_domain: null
132
+ tech_vocabulary: null
133
+ risk:
134
+ contamination: none
135
+ overlap_check:
136
+ done: true
137
+ pointers_flagged: 0
138
+ pointers_removed: 0
139
+ by_class: {}
140
+ benches: {}
141
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
142
+ pii: null
143
+ pii_evidence: null
144
+ malware: null
145
+ ship_bytes: unknown
146
+ subset:
147
+ rule: "take all rows; licence, size and version are labels, not filters"
148
+ rows_in_subset: 8665027
149
+ last_built: null
150
+ access:
151
+ method: "GET uri (LogDog, raw text log, anonymous). Keep to about one request per second per host."
152
+ verify: "No size or hash is published: size and hashes are computed when the file is downloaded (hash_source computed_on_fetch). The stored file is the raw text log."
153
+ hosts: {"https://logs.chromium.org": 8665027}
154
+ uri_example: "https://logs.chromium.org/logs/chromium/buildbucket/cr-buildbucket/8669103391362558593/+/u/blink_python_tests_on_Mac-13/stdout?format=raw"
155
+ needs_auth: {"false": 8665027}
156
+ credential: null
157
+ hash_source: {computed_on_fetch: 8665027}
158
+ range_supported: null
159
+ reachable:
160
+ index_flag: {unknown: 8665027}
161
+ index_check: null
162
+ our_sample: null
163
+ drift: null
164
+ sample: null
165
+ evidence:
166
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
167
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
168
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
169
+ class: {fill: unfilled, basis: "ships not judged"}
170
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
171
+ form.by_ext:
172
+ fill: auto
173
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
174
+ checked_at: "2026-10-03"
175
+ form.date_years:
176
+ fill: auto
177
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
178
+ checked_at: "2026-10-03"
179
+ form.over_cap:
180
+ fill: auto
181
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
182
+ checked_at: "2026-10-03"
183
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
184
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ form.size_quantiles:
186
+ fill: auto
187
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
188
+ checked_at: "2026-10-03"
189
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
190
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
191
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
192
+ provenance.licence_tiers:
193
+ fill: auto
194
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
195
+ checked_at: "2026-10-03"
196
+ provenance.spdx_top:
197
+ fill: auto
198
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
199
+ checked_at: "2026-10-03"
200
+ risk.contamination:
201
+ fill: auto
202
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
203
+ checked_at: "2026-10-03"
204
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
208
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
212
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
213
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
214
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ snapshot.rev:
221
+ fill: auto
222
+ basis: "last commit listed in the publish receipt of this source"
223
+ checked_at: "2026-10-03"
224
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
228
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ subset:
235
+ fill: auto
236
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
237
+ checked_at: "2026-10-03"
238
+ unit.pointer_count:
239
+ fill: auto
240
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
241
+ checked_at: "2026-10-03"
242
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
243
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
244
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/common-crawl.yaml ADDED
@@ -0,0 +1,1344 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: common-crawl
8
+ parent: null
9
+ title: "Common Crawl"
10
+ description: "Office and document files (including ICS, EML and MBOX) captured by Common Crawl in crawl CC-MAIN-2026-30. Each pointer is one HTTP response record inside a WARC file, retrieved by byte range; the hash covers the HTTP response body. Payloads are deduplicated within the crawl by hash. Common Crawl truncates each response at 1 MiB; truncated records are marked in source_meta.content_truncated."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: "0c3e02b5d554acb761c74328c0cbc10d80dffc10", at: "2026-10-03"}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "Common Crawl Terms of Use"
17
+ licence_url: "https://commoncrawl.org/terms-of-use"
18
+ licence_statement: "Each page remains under the copyright of its owner. The Terms of Use say: \"YOU AGREE TO RESPECT THE COPYRIGHTS ... OF THIRD PARTIES\"."
19
+ per_collection: null
20
+ training_ok: unknown
21
+ redistribution_ok: unknown
22
+ attribution: null
23
+ per_file: null
24
+ licence_tiers: {unknown/unknown: 50973470}
25
+ spdx_top: {"(none)": 50973470}
26
+ spdx_distinct: 1
27
+ licence_sample_check:
28
+ sample_n: 30
29
+ declared: 0
30
+ unknown_or_missing: 30
31
+ note: "each page keeps its owner's copyright; Common Crawl terms apply only to the crawl"
32
+ unit:
33
+ pointer_level: record
34
+ pointer_count: 50973470
35
+ units: {"WARC record": 50973470}
36
+ state: expanded
37
+ expanded_count: null
38
+ bytes_total: 39894487711582
39
+ rows_without_size: 0
40
+ collections:
41
+ distinct: 2
42
+ top: {CC-MAIN-2026-30: 24898464, CC-MAIN-2026-34: 26075006}
43
+ sub_collections: [CC-MAIN-2026-30, CC-MAIN-2026-34, CC-MAIN-2026-39]
44
+ batches: null
45
+ n_ext: 22
46
+ usable_unit: "one document file (the HTTP body of one crawled response)"
47
+ rows_not_counted:
48
+ shards: 493
49
+ rows: 24633336
50
+ note: "listed in the collection header but not present where the cards were computed; counts above leave them out"
51
+ form:
52
+ by_ext:
53
+ pdf: 41172129
54
+ txt: 5005937
55
+ xml: 2208532
56
+ ics: 1874275
57
+ docx: 143730
58
+ csv: 112578
59
+ doc: 103307
60
+ xlsx: 71873
61
+ rtf: 55526
62
+ odt: 51730
63
+ mbox: 49306
64
+ xls: 40831
65
+ eml: 32968
66
+ ods: 21546
67
+ pptx: 10358
68
+ xlsm: 8609
69
+ ppt: 5483
70
+ odp: 2321
71
+ docm: 869
72
+ odf: 648
73
+ "(2 other types)": 914
74
+ bytes_p50: 223235
75
+ bytes_p95: 4612860
76
+ bytes_max: 5253409
77
+ size_quantiles:
78
+ max: 5253409
79
+ min: 441
80
+ p25: 54245
81
+ p50: 223235
82
+ p75: 751565
83
+ p90: 2621144
84
+ p95: 4612860
85
+ p99: 5207962
86
+ p99.9: 5243854
87
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
88
+ by_ext_size:
89
+ - {ext: pdf, rows: 41172129, bytes: 39609542380625, p50: 338652, p95: 4855584, max: 5253409}
90
+ - {ext: txt, rows: 5005937, bytes: 119550684269, p50: 1971, p95: 56549, max: 5245905}
91
+ - {ext: xml, rows: 2208532, bytes: 30366684237, p50: 2840, p95: 37509, max: 3988019}
92
+ - {ext: ics, rows: 1874275, bytes: 3584989942, p50: 1385, p95: 2617, max: 3870185}
93
+ - {ext: docx, rows: 143730, bytes: 32081514797, p50: 36623, p95: 1007832, max: 5251002}
94
+ - {ext: csv, rows: 112578, bytes: 2477883002, p50: 1748, p95: 71136, max: 3950412}
95
+ - {ext: doc, rows: 103307, bytes: 14805787514, p50: 24032, p95: 539682, max: 5245487}
96
+ - {ext: xlsx, rows: 71873, bytes: 15397887294, p50: 20185, p95: 1076462, max: 5245898}
97
+ - {ext: rtf, rows: 55526, bytes: 1168115467, p50: 6904, p95: 65249, max: 5244134}
98
+ - {ext: odt, rows: 51730, bytes: 9868516071, p50: 23824, p95: 760438, max: 5245750}
99
+ - {ext: mbox, rows: 49306, bytes: 1368477352, p50: 3188, p95: 56760, max: 4550079}
100
+ - {ext: xls, rows: 40831, bytes: 3599937938, p50: 15805, p95: 328066, max: 5244030}
101
+ - {ext: eml, rows: 32968, bytes: 1816485170, p50: 3020, p95: 253776, max: 5194801}
102
+ - {ext: ods, rows: 21546, bytes: 3429880837, p50: 21992, p95: 520841, max: 5245722}
103
+ - {ext: pptx, rows: 10358, bytes: 27063863057, p50: 2211886, p95: 5218609, max: 5245837}
104
+ - {ext: xlsm, rows: 8609, bytes: 6713128639, p50: 272043, p95: 3830892, max: 5243626}
105
+ - {ext: ppt, rows: 5483, bytes: 6362983268, p50: 299803, p95: 5124168, max: 5244904}
106
+ - {ext: odp, rows: 2321, bytes: 3721323066, p50: 644018, p95: 5216134, max: 5245488}
107
+ - {ext: docm, rows: 869, bytes: 260592042, p50: 68837, p95: 1276383, max: 5244060}
108
+ - {ext: odf, rows: 648, bytes: 87398181, p50: 7576, p95: 535388, max: 5161144}
109
+ over_cap:
110
+ cap: 61944279
111
+ rows_over: 0
112
+ bytes_over: 0
113
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
114
+ largest:
115
+ - pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260722221328-20260723011328-00446.warc.gz@743785371"
116
+ size_bytes: 5253409
117
+ - pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260807175542-20260807205542-00668.warc.gz@179336775"
118
+ size_bytes: 5251002
119
+ - pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260807175542-20260807205542-00024.warc.gz@171902258"
120
+ size_bytes: 5250980
121
+ - pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260808060658-20260808090658-00771.warc.gz@132416122"
122
+ size_bytes: 5248886
123
+ - pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260720151718-20260720181718-00122.warc.gz@707279248"
124
+ size_bytes: 5248818
125
+ date_years:
126
+ key: origin.captured_at
127
+ rows_with_date: 50973470
128
+ counts: {"2026": 50973470}
129
+ media_type:
130
+ application/msword: 103234
131
+ application/pdf: 41172129
132
+ application/rtf: 55526
133
+ application/vnd.oasis.opendocument.text: 51690
134
+ application/vnd.openxmlformats-officedocument.spreadsheetml.sheet: 71873
135
+ application/vnd.openxmlformats-officedocument.wordprocessingml.document: 143730
136
+ application/xml: 2208521
137
+ text/calendar: 1874275
138
+ text/csv: 112294
139
+ text/plain: 5005937
140
+ facets:
141
+ content_truncated: {disconnect: 25854, length: 3040960, time: 65067}
142
+ artifact: [documents-text, structured-data]
143
+ source_origin: []
144
+ parse_cost: needs_tool
145
+ parse_quirks:
146
+ - "many languages and scripts (Spanish, French, Galician, Portuguese, Japanese, Russian, Ukrainian, English)"
147
+ - "scanned PDFs with no text layer"
148
+ - "bodies cut by the crawler at 5 MiB are unreadable"
149
+ - "wrong extension: RIS, crash logs and documentation dumps arrive as .txt"
150
+ structure:
151
+ self_describing: false
152
+ integrity_builtin: []
153
+ has_reference_output: false
154
+ invariants: [sha1_matches_http_body, pdf_xref_resolves_unless_truncated]
155
+ tools: [pdftotext, pdfinfo, libreoffice, file, warcio]
156
+ ships:
157
+ problem_statement: {present: false}
158
+ reference_solution:
159
+ share: {k: 0, "n": 30}
160
+ tests: {present: false, coverage: none, runs: null}
161
+ env_definition: {present: false, kind: none, runs: null}
162
+ known_answer: {present: false}
163
+ generator: {present: false}
164
+ upstream_labels:
165
+ present: true
166
+ fields: [origin.url, origin.captured_at, media_type, source_meta.content_truncated]
167
+ class: raw
168
+ subject:
169
+ domain_span: [D3.5, D2.6]
170
+ shape_span: [S4_data_to_artifact, S7_non_code_artifact]
171
+ skill_family_span: [skill.data, skill.media]
172
+ ont_domain: [language-documents, general]
173
+ tech_vocabulary: [WARC, PDF, OLE2, OOXML, RTF, iCalendar, RIS, OAI-PMH, sitemap]
174
+ risk:
175
+ contamination: tb_overlap
176
+ overlap_check:
177
+ done: true
178
+ pointers_flagged: 57
179
+ pointers_removed: 18
180
+ by_class: {data_address: 3, generic_text: 7, task_file: 15, trivial_small: 32}
181
+ benches:
182
+ APEX-Agents: 2
183
+ BrowseComp-Plus: 2
184
+ "FrontierSWE v2": 13
185
+ GDPval: 1
186
+ "MCP Atlas": 1
187
+ NL2Repo: 24
188
+ "OSWorld 2": 11
189
+ "SWE-bench Multimodal": 1
190
+ Terminal-Bench: 7
191
+ WideSearch: 1
192
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
193
+ pii: present
194
+ pii_evidence:
195
+ sample_n: 30
196
+ files_with_email: 2
197
+ files_with_named_individuals: 4
198
+ note: "grades next to names and masked ID numbers in #09"
199
+ malware: null
200
+ ship_bytes: unknown
201
+ subset:
202
+ rule: "take all rows; licence, size and version are labels, not filters"
203
+ rows_in_subset: 50973470
204
+ last_built:
205
+ date: "2026-10-02"
206
+ rule: "older rule that filtered by licence tier"
207
+ rows: 24633336
208
+ rows_in_subset: 0
209
+ excluded_by_licence: 24633336
210
+ duplicates: 0
211
+ access:
212
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
213
+ verify: "Request the byte range, which is one gzip-compressed WARC (or ARC) record; strip the WARC and HTTP headers and undo chunked transfer encoding, then compare the sha1 of the HTTP body with hashes.sha1 (hash_scope http_payload). The stored file is the HTTP body."
214
+ hosts: {"s3://commoncrawl": 50973470}
215
+ uri_example: "s3://commoncrawl/crawl-data/CC-MAIN-2026-30/segments/1783663951123.52/warc/CC-MAIN-20260710070534-20260710100534-00000.warc.gz"
216
+ needs_auth: {"false": 50973470}
217
+ credential: none
218
+ hash_source: {source_published: 50973470}
219
+ range_supported: null
220
+ reachable:
221
+ index_flag: {unknown: 50973470}
222
+ index_check: null
223
+ our_sample: {checked_at: "2026-10-04", n_tried: 30, n_ok: 30, hash_checked: 30}
224
+ drift: "stable: a WARC file is never rewritten once a crawl is published; the origin page may change, but the pointer names the captured copy"
225
+ sample:
226
+ method: "strata are file type times the size quartile of this source (q1 to q4 by its own p25, p50, p75; unk for rows without a size); half of the samples go one per stratum to the largest strata, the rest in proportion to stratum size (largest remainder); within a stratum the rows with the smallest sha256(seed:pointer_id) are taken, so the draw is reproducible from the index"
227
+ seed: 20261004
228
+ "n": 30
229
+ n_drawn: 30
230
+ n_read: 30
231
+ unread: []
232
+ rule_n: "10 when one file type holds at least 95 percent of rows, else 30"
233
+ size_quartiles: [54245, 223235, 751565]
234
+ size_filter: "only files of at most 15000000 bytes are candidates"
235
+ strata_candidates:
236
+ "pdf|q4": 12665418
237
+ "pdf|q3": 12618602
238
+ "pdf|q2": 12442981
239
+ "txt|q1": 4750637
240
+ "pdf|q1": 3445128
241
+ "xml|q1": 2142163
242
+ "ics|q1": 1871767
243
+ "txt|q2": 146122
244
+ "csv|q1": 106288
245
+ "docx|q1": 87791
246
+ "txt|q3": 71824
247
+ "doc|q1": 70922
248
+ "rtf|q1": 52229
249
+ "xlsx|q1": 50597
250
+ "xml|q2": 49843
251
+ "mbox|q1": 46727
252
+ "odt|q1": 38834
253
+ "txt|q4": 37354
254
+ "docx|q2": 35469
255
+ "xls|q1": 30422
256
+ "eml|q1": 29231
257
+ "doc|q2": 22223
258
+ "ods|q1": 16616
259
+ "xml|q3": 12371
260
+ "xlsx|q2": 11536
261
+ "docx|q3": 11518
262
+ "docx|q4": 8952
263
+ "odt|q2": 7553
264
+ "pptx|q4": 7384
265
+ "xls|q2": 6760
266
+ "doc|q3": 6342
267
+ "xlsx|q3": 5250
268
+ "xlsx|q4": 4490
269
+ "xml|q4": 4155
270
+ "doc|q4": 3820
271
+ "csv|q2": 3184
272
+ "ods|q2": 3184
273
+ "odt|q3": 2730
274
+ "xls|q3": 2718
275
+ "odt|q4": 2613
276
+ strata_total: 88
277
+ strata_taken:
278
+ "csv|q1": 106288
279
+ "docx|q1": 87791
280
+ "doc|q1": 70922
281
+ "ics|q1": 1871767
282
+ "pdf|q1": 3445128
283
+ "pdf|q2": 12442981
284
+ "pdf|q3": 12618602
285
+ "pdf|q4": 12665418
286
+ "rtf|q1": 52229
287
+ "txt|q1": 4750637
288
+ "txt|q2": 146122
289
+ "txt|q3": 71824
290
+ "xlsx|q1": 50597
291
+ "xml|q1": 2142163
292
+ "xml|q2": 49843
293
+ fetched_at: "2026-10-04"
294
+ replaced: 0
295
+ kept: false
296
+ kept_note: "sample bytes were deleted after reading; refetch by pointer_id and compare sha256"
297
+ files:
298
+ - "no": "#00"
299
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260810221116-20260811011116-00621.warc.gz@613983620"
300
+ stratum: "csv|q1"
301
+ size_bytes: 3154
302
+ sha256: c4c32ab17f74a69026885ff2890da9dec2ceabb561b68bf533711738549e76f3
303
+ - "no": "#01"
304
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260717103357-20260717133357-00446.warc.gz@885012444"
305
+ stratum: "docx|q1"
306
+ size_bytes: 17897
307
+ sha256: c9a00a853485bf81052ca0305c5563ea045a4c1730eb5d656b8820af03440600
308
+ - "no": "#02"
309
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260720182018-20260720212018-00695.warc.gz@266594559"
310
+ stratum: "doc|q1"
311
+ size_bytes: 32768
312
+ sha256: a1e00e2974cb555187fb5e10d61c1de388c7cdc5794f2132381fe1273ec8562d
313
+ - "no": "#03"
314
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260722100027-20260722130027-00531.warc.gz@358355136"
315
+ stratum: "ics|q1"
316
+ size_bytes: 755
317
+ sha256: "4cea0bc812b57c7dab20d3c1f39fb990f71a0878de583319dd2959f7f08fd4fb"
318
+ - "no": "#04"
319
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260720030358-20260720060358-00030.warc.gz@188078856"
320
+ stratum: "pdf|q1"
321
+ size_bytes: 24278
322
+ sha256: dd212cebcb6aee14ce4b04d64c32cff3057003de1543c443ce99e4b0b6cd958a
323
+ - "no": "#05"
324
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260811011751-20260811041751-00201.warc.gz@395729629"
325
+ stratum: "pdf|q1"
326
+ size_bytes: 44810
327
+ sha256: b505ed97ce97e7df9699c72d4e3fc905e99b3569f162641abd81114d1eb298e2
328
+ - "no": "#06"
329
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260808030320-20260808060320-00372.warc.gz@152337648"
330
+ stratum: "pdf|q2"
331
+ size_bytes: 179017
332
+ sha256: bf404e7971d24d95e4bf9a99f89653775d3eb72057860f27d5ec6b3f7d2bec21
333
+ - "no": "#07"
334
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260812230232-20260813020232-00192.warc.gz@585749941"
335
+ stratum: "pdf|q2"
336
+ size_bytes: 102614
337
+ sha256: "217eeff3a64e8209d0274a4bbaa09a1c3c3176c935c5e9d637d3946c30bacc4c"
338
+ - "no": "#08"
339
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260814144954-20260814174954-00280.warc.gz@305972432"
340
+ stratum: "pdf|q2"
341
+ size_bytes: 88218
342
+ sha256: "47aa628dd3ecbedc2c541973c2d61bc45cc64d2c792b4de06f5c66deb50f461c"
343
+ - "no": "#09"
344
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260718050035-20260718080035-00357.warc.gz@591576795"
345
+ stratum: "pdf|q2"
346
+ size_bytes: 158480
347
+ sha256: f4941cea60ef27873a8b3211dafc779ff1e16cc888a48883a2d1a3a60a942a69
348
+ - "no": "#10"
349
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260817004738-20260817034738-00347.warc.gz@391140057"
350
+ stratum: "pdf|q2"
351
+ size_bytes: 267000
352
+ sha256: ca5effec134cdd05d7c2d568c71c7828417a429bc30784d10d124f7947e29cfa
353
+ - "no": "#11"
354
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260713205422-20260713235422-00507.warc.gz@557119412"
355
+ stratum: "pdf|q3"
356
+ size_bytes: 546401
357
+ sha256: bfa38f236246867cb2043a0bff46ed2fdf2b460816b575f5584997db02eec197
358
+ - "no": "#12"
359
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260817065350-20260817095350-00696.warc.gz@545992600"
360
+ stratum: "pdf|q3"
361
+ size_bytes: 535703
362
+ sha256: dadea7d64b81b830284302d7f246d4966e3bc2eceba396463fd65a21d6b7d324
363
+ - "no": "#13"
364
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260721002628-20260721032628-00312.warc.gz@140433833"
365
+ stratum: "pdf|q3"
366
+ size_bytes: 300628
367
+ sha256: "7f9f3ca5174e9bae30684a7eceaea23ff6d99dd44c158dfed1bb66abd9bafbac"
368
+ - "no": "#14"
369
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260808090928-20260808120928-00866.warc.gz@217620612"
370
+ stratum: "pdf|q3"
371
+ size_bytes: 362445
372
+ sha256: "115ed8bb8fc576deb7a4eb63cba5de24da87dfc6467645fa04fc73f49faf5e13"
373
+ - "no": "#15"
374
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260715063210-20260715093210-00086.warc.gz@719568456"
375
+ stratum: "pdf|q3"
376
+ size_bytes: 672282
377
+ sha256: e666fbca3d49f02e84e6b46a0eb46911824cbd7798868f61d4e887324bdb13da
378
+ - "no": "#16"
379
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260819134558-20260819164558-00981.warc.gz@602003197"
380
+ stratum: "pdf|q4"
381
+ size_bytes: 5242880
382
+ sha256: "4b201abf08bb0300c20a50a787efd71c2dd31ab2e5b485940a56dcbd52819cec"
383
+ - "no": "#17"
384
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260718050035-20260718080035-00118.warc.gz@204156968"
385
+ stratum: "pdf|q4"
386
+ size_bytes: 1997538
387
+ sha256: "87659139d41010181812ba389233fb5394f55b1c5a55d6f3d1e37e8c686896f0"
388
+ - "no": "#18"
389
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260808212239-20260809002239-00251.warc.gz@11600015"
390
+ stratum: "pdf|q4"
391
+ size_bytes: 2306900
392
+ sha256: a6dfcf139e007767bf9623840aacf0397ab6cb342cff88a27cabc38d3b069ab1
393
+ - "no": "#19"
394
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260721154245-20260721184245-00004.warc.gz@815523966"
395
+ stratum: "pdf|q4"
396
+ size_bytes: 1972337
397
+ sha256: "768e3264570a0a3e6d9c2d0df4e4c6cdef483976d53be8ad35d845556bb34e1c"
398
+ - "no": "#20"
399
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260812014208-20260812044208-00441.warc.gz@945475682"
400
+ stratum: "pdf|q4"
401
+ size_bytes: 1191811
402
+ sha256: bf5811a93643a60631b50eef5d77f46b0b07fdf4cf171244fe59a3b20678370d
403
+ - "no": "#21"
404
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260721002628-20260721032628-00478.warc.gz@18749819"
405
+ stratum: "rtf|q1"
406
+ size_bytes: 159102
407
+ sha256: ab55c18f59ae5e6b1eca3180b0ef6f944304b177e000c61875adabb4d1a62aac
408
+ - "no": "#22"
409
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260715093621-20260715123621-00063.warc.gz@377006069"
410
+ stratum: "txt|q1"
411
+ size_bytes: 754
412
+ sha256: "89483582c2a9a8f1699078e5413c0c9c5db7a6a72d50338816a389dd263d3445"
413
+ - "no": "#23"
414
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260814235839-20260815025839-00644.warc.gz@259368549"
415
+ stratum: "txt|q1"
416
+ size_bytes: 538
417
+ sha256: "2b6b58e028b2ed02c1be110a73e4cc5d1268a760fe615b578b6f478ea6744c64"
418
+ - "no": "#24"
419
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260811102717-20260811132717-00733.warc.gz@461558458"
420
+ stratum: "txt|q2"
421
+ size_bytes: 368408
422
+ sha256: b1ff5e302ceb32c1c9b1721ee1a64474fa4c755b4f91d374c25b34229fe087bf
423
+ - "no": "#25"
424
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260813202929-20260813232929-00183.warc.gz@200708294"
425
+ stratum: "txt|q3"
426
+ size_bytes: 2312714
427
+ sha256: b6662484a919f89e705bf6e40e39f8664f1be677ab6f598f577603d8203b1bdd
428
+ - "no": "#26"
429
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260817191046-20260817221046-00703.warc.gz@664541304"
430
+ stratum: "xlsx|q1"
431
+ size_bytes: 57280
432
+ sha256: "09b1a9b5b78fbce112e2233b56e368d6286652070719337d86aff50dd971d6fb"
433
+ - "no": "#27"
434
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260720030358-20260720060358-00784.warc.gz@249092754"
435
+ stratum: "xml|q1"
436
+ size_bytes: 2899
437
+ sha256: "8cbcd11ba8ce9e3f5e291745127de98a0108d87363b9ca80b6654129f6a96443"
438
+ - "no": "#28"
439
+ pointer_id: "common-crawl:CC-MAIN-2026-34:CC-MAIN-20260808212239-20260809002239-00481.warc.gz@256395155"
440
+ stratum: "xml|q1"
441
+ size_bytes: 5170
442
+ sha256: b954dcf99d08bd7c2b414240516ca5b24100d869ff4267f33fde6857b1bd2886
443
+ - "no": "#29"
444
+ pointer_id: "common-crawl:CC-MAIN-2026-30:CC-MAIN-20260718141315-20260718171315-00542.warc.gz@8285868"
445
+ stratum: "xml|q2"
446
+ size_bytes: 3363736
447
+ sha256: a532b89db28bbd0f40858b1016c2a16b435c61f645ef6e008ce8429913d44375
448
+ evidence:
449
+ access.credential:
450
+ fill: judgment
451
+ basis: "all 30 fetched by byte range from data.commoncrawl.org without a key"
452
+ files:
453
+ - "#00"
454
+ - "#01"
455
+ - "#02"
456
+ - "#03"
457
+ - "#04"
458
+ - "#05"
459
+ - "#06"
460
+ - "#07"
461
+ - "#08"
462
+ - "#09"
463
+ - "#10"
464
+ - "#11"
465
+ - "#12"
466
+ - "#13"
467
+ - "#14"
468
+ - "#15"
469
+ - "#16"
470
+ - "#17"
471
+ - "#18"
472
+ - "#19"
473
+ - "#20"
474
+ - "#21"
475
+ - "#22"
476
+ - "#23"
477
+ - "#24"
478
+ - "#25"
479
+ - "#26"
480
+ - "#27"
481
+ - "#28"
482
+ - "#29"
483
+ sample_n: 30
484
+ checked_at: "2026-10-04"
485
+ access.drift:
486
+ fill: judgment
487
+ basis: "every pointer carries crawl id, WARC file and offset"
488
+ files:
489
+ - "#00"
490
+ - "#01"
491
+ - "#02"
492
+ - "#03"
493
+ - "#04"
494
+ - "#05"
495
+ - "#06"
496
+ - "#07"
497
+ - "#08"
498
+ - "#09"
499
+ - "#10"
500
+ - "#11"
501
+ - "#12"
502
+ - "#13"
503
+ - "#14"
504
+ - "#15"
505
+ - "#16"
506
+ - "#17"
507
+ - "#18"
508
+ - "#19"
509
+ - "#20"
510
+ - "#21"
511
+ - "#22"
512
+ - "#23"
513
+ - "#24"
514
+ - "#25"
515
+ - "#26"
516
+ - "#27"
517
+ - "#28"
518
+ - "#29"
519
+ sample_n: 30
520
+ checked_at: "2026-10-04"
521
+ access.range_supported: {fill: unfilled, basis: "no sample evidence given for this field"}
522
+ class:
523
+ fill: derived
524
+ basis: "rule over ships.tests, env_definition, known_answer, generator and structure; manifest_not_run when tests and a dependency manifest are present but the build was not run"
525
+ form.artifact:
526
+ fill: judgment
527
+ basis: "ruling after independent re-judge: the one crash log (#24) of 30 samples is not central, so logs-traces is dropped; documents in #01, #02, #04 to #21, #25 and records in #00, #03, #22, #23, #26 to #29"
528
+ files:
529
+ - "#00"
530
+ - "#01"
531
+ - "#02"
532
+ - "#03"
533
+ - "#04"
534
+ - "#05"
535
+ - "#06"
536
+ - "#07"
537
+ - "#08"
538
+ - "#09"
539
+ - "#10"
540
+ - "#11"
541
+ - "#12"
542
+ - "#13"
543
+ - "#14"
544
+ - "#15"
545
+ - "#16"
546
+ - "#17"
547
+ - "#18"
548
+ - "#19"
549
+ - "#20"
550
+ - "#21"
551
+ - "#22"
552
+ - "#23"
553
+ - "#24"
554
+ - "#25"
555
+ - "#26"
556
+ - "#27"
557
+ - "#28"
558
+ - "#29"
559
+ sample_n: 30
560
+ checked_at: "2026-10-04"
561
+ form.by_ext:
562
+ fill: auto
563
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
564
+ checked_at: "2026-10-03"
565
+ form.date_years:
566
+ fill: auto
567
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
568
+ checked_at: "2026-10-03"
569
+ form.over_cap:
570
+ fill: auto
571
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
572
+ checked_at: "2026-10-03"
573
+ form.parse_cost:
574
+ fill: judgment
575
+ basis: "PDF, OLE2 DOC, OOXML and RTF need converters (pdftotext, file(1) reports binary containers for #01, #02, #26); only the CSV, TXT, ICS and XML samples are plain text"
576
+ files: ["#01", "#02", "#04", "#21", "#26"]
577
+ sample_n: 30
578
+ checked_at: "2026-10-04"
579
+ form.parse_quirks:
580
+ fill: judgment
581
+ basis: "pdftotext returns 0 chars for #10, #11, #19, #20; #16 is exactly 5242880 bytes, marked content_truncated=length, and pdfinfo cannot find the xref table; #22 and #23 are RIS citations, #24 a syzkaller log, #25 an llms-full.txt dump"
582
+ files: ["#09", "#10", "#11", "#12", "#16", "#19", "#20", "#21", "#22", "#24", "#25"]
583
+ sample_n: 30
584
+ checked_at: "2026-10-04"
585
+ form.size_quantiles:
586
+ fill: auto
587
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
588
+ checked_at: "2026-10-03"
589
+ form.source_origin:
590
+ fill: judgment
591
+ basis: "no source code in the sample"
592
+ files:
593
+ - "#00"
594
+ - "#01"
595
+ - "#02"
596
+ - "#03"
597
+ - "#04"
598
+ - "#05"
599
+ - "#06"
600
+ - "#07"
601
+ - "#08"
602
+ - "#09"
603
+ - "#10"
604
+ - "#11"
605
+ - "#12"
606
+ - "#13"
607
+ - "#14"
608
+ - "#15"
609
+ - "#16"
610
+ - "#17"
611
+ - "#18"
612
+ - "#19"
613
+ - "#20"
614
+ - "#21"
615
+ - "#22"
616
+ - "#23"
617
+ - "#24"
618
+ - "#25"
619
+ - "#26"
620
+ - "#27"
621
+ - "#28"
622
+ - "#29"
623
+ sample_n: 30
624
+ checked_at: "2026-10-04"
625
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
626
+ provenance.licence_sample_check:
627
+ fill: judgment
628
+ basis: "no licence statement found in the text heads"
629
+ files:
630
+ - "#00"
631
+ - "#01"
632
+ - "#02"
633
+ - "#03"
634
+ - "#04"
635
+ - "#05"
636
+ - "#06"
637
+ - "#07"
638
+ - "#08"
639
+ - "#09"
640
+ - "#10"
641
+ - "#11"
642
+ - "#12"
643
+ - "#13"
644
+ - "#14"
645
+ - "#15"
646
+ - "#16"
647
+ - "#17"
648
+ - "#18"
649
+ - "#19"
650
+ - "#20"
651
+ - "#21"
652
+ - "#22"
653
+ - "#23"
654
+ - "#24"
655
+ - "#25"
656
+ - "#26"
657
+ - "#27"
658
+ - "#28"
659
+ - "#29"
660
+ sample_n: 30
661
+ checked_at: "2026-10-04"
662
+ provenance.licence_tiers:
663
+ fill: auto
664
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
665
+ checked_at: "2026-10-03"
666
+ provenance.spdx_top:
667
+ fill: auto
668
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
669
+ checked_at: "2026-10-03"
670
+ risk.contamination:
671
+ fill: auto
672
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
673
+ checked_at: "2026-10-03"
674
+ risk.malware: {fill: unfilled, basis: "no sample evidence given for this field"}
675
+ risk.pii:
676
+ fill: judgment
677
+ basis: "#09 lists named candidates with partly masked national ID numbers and exam grades; #06 is a court order with attorney emails; #13 carries a personal gmail address; #02 and #14 name private persons; re-checked under the WP-T-44c definition (a user name on a platform counts as present)"
678
+ files: ["#02", "#06", "#09", "#13", "#14"]
679
+ sample_n: 30
680
+ checked_at: "2026-10-04"
681
+ risk.pii_evidence:
682
+ fill: judgment
683
+ basis: "email regex over the digests plus reading the text heads"
684
+ files: ["#02", "#06", "#09", "#13", "#14"]
685
+ sample_n: 30
686
+ checked_at: "2026-10-04"
687
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
688
+ ships.env_definition.kind:
689
+ fill: judgment
690
+ basis: "no environment definition"
691
+ files:
692
+ - "#00"
693
+ - "#01"
694
+ - "#02"
695
+ - "#03"
696
+ - "#04"
697
+ - "#05"
698
+ - "#06"
699
+ - "#07"
700
+ - "#08"
701
+ - "#09"
702
+ - "#10"
703
+ - "#11"
704
+ - "#12"
705
+ - "#13"
706
+ - "#14"
707
+ - "#15"
708
+ - "#16"
709
+ - "#17"
710
+ - "#18"
711
+ - "#19"
712
+ - "#20"
713
+ - "#21"
714
+ - "#22"
715
+ - "#23"
716
+ - "#24"
717
+ - "#25"
718
+ - "#26"
719
+ - "#27"
720
+ - "#28"
721
+ - "#29"
722
+ sample_n: 30
723
+ checked_at: "2026-10-04"
724
+ ships.env_definition.present:
725
+ fill: judgment
726
+ basis: "no environment definition"
727
+ files:
728
+ - "#00"
729
+ - "#01"
730
+ - "#02"
731
+ - "#03"
732
+ - "#04"
733
+ - "#05"
734
+ - "#06"
735
+ - "#07"
736
+ - "#08"
737
+ - "#09"
738
+ - "#10"
739
+ - "#11"
740
+ - "#12"
741
+ - "#13"
742
+ - "#14"
743
+ - "#15"
744
+ - "#16"
745
+ - "#17"
746
+ - "#18"
747
+ - "#19"
748
+ - "#20"
749
+ - "#21"
750
+ - "#22"
751
+ - "#23"
752
+ - "#24"
753
+ - "#25"
754
+ - "#26"
755
+ - "#27"
756
+ - "#28"
757
+ - "#29"
758
+ sample_n: 30
759
+ checked_at: "2026-10-04"
760
+ ships.env_definition.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
761
+ ships.generator.present:
762
+ fill: judgment
763
+ basis: "no generator"
764
+ files:
765
+ - "#00"
766
+ - "#01"
767
+ - "#02"
768
+ - "#03"
769
+ - "#04"
770
+ - "#05"
771
+ - "#06"
772
+ - "#07"
773
+ - "#08"
774
+ - "#09"
775
+ - "#10"
776
+ - "#11"
777
+ - "#12"
778
+ - "#13"
779
+ - "#14"
780
+ - "#15"
781
+ - "#16"
782
+ - "#17"
783
+ - "#18"
784
+ - "#19"
785
+ - "#20"
786
+ - "#21"
787
+ - "#22"
788
+ - "#23"
789
+ - "#24"
790
+ - "#25"
791
+ - "#26"
792
+ - "#27"
793
+ - "#28"
794
+ - "#29"
795
+ sample_n: 30
796
+ checked_at: "2026-10-04"
797
+ ships.known_answer.present:
798
+ fill: judgment
799
+ basis: "no answer field inside the documents"
800
+ files:
801
+ - "#00"
802
+ - "#01"
803
+ - "#02"
804
+ - "#03"
805
+ - "#04"
806
+ - "#05"
807
+ - "#06"
808
+ - "#07"
809
+ - "#08"
810
+ - "#09"
811
+ - "#10"
812
+ - "#11"
813
+ - "#12"
814
+ - "#13"
815
+ - "#14"
816
+ - "#15"
817
+ - "#16"
818
+ - "#17"
819
+ - "#18"
820
+ - "#19"
821
+ - "#20"
822
+ - "#21"
823
+ - "#22"
824
+ - "#23"
825
+ - "#24"
826
+ - "#25"
827
+ - "#26"
828
+ - "#27"
829
+ - "#28"
830
+ - "#29"
831
+ sample_n: 30
832
+ checked_at: "2026-10-04"
833
+ ships.problem_statement.present:
834
+ fill: judgment
835
+ basis: "crawled documents, not tasks"
836
+ files:
837
+ - "#00"
838
+ - "#01"
839
+ - "#02"
840
+ - "#03"
841
+ - "#04"
842
+ - "#05"
843
+ - "#06"
844
+ - "#07"
845
+ - "#08"
846
+ - "#09"
847
+ - "#10"
848
+ - "#11"
849
+ - "#12"
850
+ - "#13"
851
+ - "#14"
852
+ - "#15"
853
+ - "#16"
854
+ - "#17"
855
+ - "#18"
856
+ - "#19"
857
+ - "#20"
858
+ - "#21"
859
+ - "#22"
860
+ - "#23"
861
+ - "#24"
862
+ - "#25"
863
+ - "#26"
864
+ - "#27"
865
+ - "#28"
866
+ - "#29"
867
+ sample_n: 30
868
+ checked_at: "2026-10-04"
869
+ ships.reference_solution.share:
870
+ fill: judgment
871
+ basis: "0 of 30 sample files read: no sample file carries a fix, patch, solved answer or worked solution"
872
+ files: []
873
+ sample_n: 30
874
+ checked_at: "2026-10-04"
875
+ ships.tests.coverage:
876
+ fill: judgment
877
+ basis: "no tests"
878
+ files:
879
+ - "#00"
880
+ - "#01"
881
+ - "#02"
882
+ - "#03"
883
+ - "#04"
884
+ - "#05"
885
+ - "#06"
886
+ - "#07"
887
+ - "#08"
888
+ - "#09"
889
+ - "#10"
890
+ - "#11"
891
+ - "#12"
892
+ - "#13"
893
+ - "#14"
894
+ - "#15"
895
+ - "#16"
896
+ - "#17"
897
+ - "#18"
898
+ - "#19"
899
+ - "#20"
900
+ - "#21"
901
+ - "#22"
902
+ - "#23"
903
+ - "#24"
904
+ - "#25"
905
+ - "#26"
906
+ - "#27"
907
+ - "#28"
908
+ - "#29"
909
+ sample_n: 30
910
+ checked_at: "2026-10-04"
911
+ ships.tests.present:
912
+ fill: judgment
913
+ basis: "no tests"
914
+ files:
915
+ - "#00"
916
+ - "#01"
917
+ - "#02"
918
+ - "#03"
919
+ - "#04"
920
+ - "#05"
921
+ - "#06"
922
+ - "#07"
923
+ - "#08"
924
+ - "#09"
925
+ - "#10"
926
+ - "#11"
927
+ - "#12"
928
+ - "#13"
929
+ - "#14"
930
+ - "#15"
931
+ - "#16"
932
+ - "#17"
933
+ - "#18"
934
+ - "#19"
935
+ - "#20"
936
+ - "#21"
937
+ - "#22"
938
+ - "#23"
939
+ - "#24"
940
+ - "#25"
941
+ - "#26"
942
+ - "#27"
943
+ - "#28"
944
+ - "#29"
945
+ sample_n: 30
946
+ checked_at: "2026-10-04"
947
+ ships.tests.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
948
+ ships.upstream_labels.fields:
949
+ fill: judgment
950
+ basis: "pointer keys copied from the CDX index"
951
+ files:
952
+ - "#00"
953
+ - "#01"
954
+ - "#02"
955
+ - "#03"
956
+ - "#04"
957
+ - "#05"
958
+ - "#06"
959
+ - "#07"
960
+ - "#08"
961
+ - "#09"
962
+ - "#10"
963
+ - "#11"
964
+ - "#12"
965
+ - "#13"
966
+ - "#14"
967
+ - "#15"
968
+ - "#16"
969
+ - "#17"
970
+ - "#18"
971
+ - "#19"
972
+ - "#20"
973
+ - "#21"
974
+ - "#22"
975
+ - "#23"
976
+ - "#24"
977
+ - "#25"
978
+ - "#26"
979
+ - "#27"
980
+ - "#28"
981
+ - "#29"
982
+ sample_n: 30
983
+ checked_at: "2026-10-04"
984
+ ships.upstream_labels.present:
985
+ fill: judgment
986
+ basis: "the crawl index records the origin URL, capture time, detected media type and a truncation flag per record"
987
+ files:
988
+ - "#00"
989
+ - "#01"
990
+ - "#02"
991
+ - "#03"
992
+ - "#04"
993
+ - "#05"
994
+ - "#06"
995
+ - "#07"
996
+ - "#08"
997
+ - "#09"
998
+ - "#10"
999
+ - "#11"
1000
+ - "#12"
1001
+ - "#13"
1002
+ - "#14"
1003
+ - "#15"
1004
+ - "#16"
1005
+ - "#17"
1006
+ - "#18"
1007
+ - "#19"
1008
+ - "#20"
1009
+ - "#21"
1010
+ - "#22"
1011
+ - "#23"
1012
+ - "#24"
1013
+ - "#25"
1014
+ - "#26"
1015
+ - "#27"
1016
+ - "#28"
1017
+ - "#29"
1018
+ sample_n: 30
1019
+ checked_at: "2026-10-04"
1020
+ snapshot.rev:
1021
+ fill: auto
1022
+ basis: "last commit listed in the publish receipt of this source"
1023
+ checked_at: "2026-10-03"
1024
+ structure.has_reference_output:
1025
+ fill: judgment
1026
+ basis: "no expected output file"
1027
+ files:
1028
+ - "#00"
1029
+ - "#01"
1030
+ - "#02"
1031
+ - "#03"
1032
+ - "#04"
1033
+ - "#05"
1034
+ - "#06"
1035
+ - "#07"
1036
+ - "#08"
1037
+ - "#09"
1038
+ - "#10"
1039
+ - "#11"
1040
+ - "#12"
1041
+ - "#13"
1042
+ - "#14"
1043
+ - "#15"
1044
+ - "#16"
1045
+ - "#17"
1046
+ - "#18"
1047
+ - "#19"
1048
+ - "#20"
1049
+ - "#21"
1050
+ - "#22"
1051
+ - "#23"
1052
+ - "#24"
1053
+ - "#25"
1054
+ - "#26"
1055
+ - "#27"
1056
+ - "#28"
1057
+ - "#29"
1058
+ sample_n: 30
1059
+ checked_at: "2026-10-04"
1060
+ structure.integrity_builtin:
1061
+ fill: judgment
1062
+ basis: "no checksum inside the documents; the sha1 belongs to the crawl index, not the file"
1063
+ files:
1064
+ - "#00"
1065
+ - "#01"
1066
+ - "#02"
1067
+ - "#03"
1068
+ - "#04"
1069
+ - "#05"
1070
+ - "#06"
1071
+ - "#07"
1072
+ - "#08"
1073
+ - "#09"
1074
+ - "#10"
1075
+ - "#11"
1076
+ - "#12"
1077
+ - "#13"
1078
+ - "#14"
1079
+ - "#15"
1080
+ - "#16"
1081
+ - "#17"
1082
+ - "#18"
1083
+ - "#19"
1084
+ - "#20"
1085
+ - "#21"
1086
+ - "#22"
1087
+ - "#23"
1088
+ - "#24"
1089
+ - "#25"
1090
+ - "#26"
1091
+ - "#27"
1092
+ - "#28"
1093
+ - "#29"
1094
+ sample_n: 30
1095
+ checked_at: "2026-10-04"
1096
+ structure.invariants:
1097
+ fill: judgment
1098
+ basis: "all 30 bodies matched the index sha1 on fetch; the only PDF whose xref fails (#16) is the one marked content_truncated"
1099
+ files: ["#16"]
1100
+ sample_n: 30
1101
+ checked_at: "2026-10-04"
1102
+ structure.self_describing:
1103
+ fill: judgment
1104
+ basis: "office and PDF bodies carry no schema; only the XML (#27, #28, #29), iCalendar (#03) and CSV header (#00) samples name their fields"
1105
+ files:
1106
+ - "#00"
1107
+ - "#01"
1108
+ - "#02"
1109
+ - "#03"
1110
+ - "#04"
1111
+ - "#05"
1112
+ - "#06"
1113
+ - "#07"
1114
+ - "#08"
1115
+ - "#09"
1116
+ - "#10"
1117
+ - "#11"
1118
+ - "#12"
1119
+ - "#13"
1120
+ - "#14"
1121
+ - "#15"
1122
+ - "#16"
1123
+ - "#17"
1124
+ - "#18"
1125
+ - "#19"
1126
+ - "#20"
1127
+ - "#21"
1128
+ - "#22"
1129
+ - "#23"
1130
+ - "#24"
1131
+ - "#25"
1132
+ - "#26"
1133
+ - "#27"
1134
+ - "#28"
1135
+ - "#29"
1136
+ sample_n: 30
1137
+ checked_at: "2026-10-04"
1138
+ structure.tools:
1139
+ fill: judgment
1140
+ basis: "the tools that opened the samples; LibreOffice for DOC, DOCX, XLSX and RTF"
1141
+ files:
1142
+ - "#00"
1143
+ - "#01"
1144
+ - "#02"
1145
+ - "#03"
1146
+ - "#04"
1147
+ - "#05"
1148
+ - "#06"
1149
+ - "#07"
1150
+ - "#08"
1151
+ - "#09"
1152
+ - "#10"
1153
+ - "#11"
1154
+ - "#12"
1155
+ - "#13"
1156
+ - "#14"
1157
+ - "#15"
1158
+ - "#16"
1159
+ - "#17"
1160
+ - "#18"
1161
+ - "#19"
1162
+ - "#20"
1163
+ - "#21"
1164
+ - "#22"
1165
+ - "#23"
1166
+ - "#24"
1167
+ - "#25"
1168
+ - "#26"
1169
+ - "#27"
1170
+ - "#28"
1171
+ - "#29"
1172
+ sample_n: 30
1173
+ checked_at: "2026-10-04"
1174
+ subject.domain_span:
1175
+ fill: judgment
1176
+ basis: "document and spreadsheet extraction is the common task (#00 to #21, #26); #24 is a kernel crash log for crash diagnosis; trimmed to at most three main values (WP-T-44c): kept the values most samples show, dropped a parent code when its subcode is kept"
1177
+ files: ["#00", "#01", "#02", "#09", "#14", "#24", "#26"]
1178
+ sample_n: 30
1179
+ checked_at: "2026-10-04"
1180
+ subject.ont_domain:
1181
+ fill: judgment
1182
+ basis: "public administration, school, product and court documents on every topic"
1183
+ files:
1184
+ - "#00"
1185
+ - "#01"
1186
+ - "#02"
1187
+ - "#03"
1188
+ - "#04"
1189
+ - "#05"
1190
+ - "#06"
1191
+ - "#07"
1192
+ - "#08"
1193
+ - "#09"
1194
+ - "#10"
1195
+ - "#11"
1196
+ - "#12"
1197
+ - "#13"
1198
+ - "#14"
1199
+ - "#15"
1200
+ - "#16"
1201
+ - "#17"
1202
+ - "#18"
1203
+ - "#19"
1204
+ - "#20"
1205
+ - "#21"
1206
+ - "#22"
1207
+ - "#23"
1208
+ - "#24"
1209
+ - "#25"
1210
+ - "#26"
1211
+ - "#27"
1212
+ - "#28"
1213
+ - "#29"
1214
+ sample_n: 30
1215
+ checked_at: "2026-10-04"
1216
+ subject.shape_span:
1217
+ fill: judgment
1218
+ basis: "documents to structured extracts (tables in #09, contract terms in #14), and documents as outputs in their own right"
1219
+ files: ["#09", "#14", "#26"]
1220
+ sample_n: 30
1221
+ checked_at: "2026-10-04"
1222
+ subject.skill_family_span:
1223
+ fill: judgment
1224
+ basis: "text and table extraction; OCR for the scanned PDFs"
1225
+ files: ["#09", "#10", "#11", "#19", "#20"]
1226
+ sample_n: 30
1227
+ checked_at: "2026-10-04"
1228
+ subject.tech_vocabulary:
1229
+ fill: judgment
1230
+ basis: "formats seen in the sample"
1231
+ files:
1232
+ - "#00"
1233
+ - "#01"
1234
+ - "#02"
1235
+ - "#03"
1236
+ - "#04"
1237
+ - "#05"
1238
+ - "#06"
1239
+ - "#07"
1240
+ - "#08"
1241
+ - "#09"
1242
+ - "#10"
1243
+ - "#11"
1244
+ - "#12"
1245
+ - "#13"
1246
+ - "#14"
1247
+ - "#15"
1248
+ - "#16"
1249
+ - "#17"
1250
+ - "#18"
1251
+ - "#19"
1252
+ - "#20"
1253
+ - "#21"
1254
+ - "#22"
1255
+ - "#23"
1256
+ - "#24"
1257
+ - "#25"
1258
+ - "#26"
1259
+ - "#27"
1260
+ - "#28"
1261
+ - "#29"
1262
+ sample_n: 30
1263
+ checked_at: "2026-10-04"
1264
+ subset:
1265
+ fill: auto
1266
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
1267
+ checked_at: "2026-10-03"
1268
+ unit.pointer_count:
1269
+ fill: auto
1270
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
1271
+ checked_at: "2026-10-03"
1272
+ unit.pointer_level:
1273
+ fill: judgment
1274
+ basis: "each pointer is one HTTP response record inside a WARC file, fetched by byte range; the stored file is the HTTP body"
1275
+ files:
1276
+ - "#00"
1277
+ - "#01"
1278
+ - "#02"
1279
+ - "#03"
1280
+ - "#04"
1281
+ - "#05"
1282
+ - "#06"
1283
+ - "#07"
1284
+ - "#08"
1285
+ - "#09"
1286
+ - "#10"
1287
+ - "#11"
1288
+ - "#12"
1289
+ - "#13"
1290
+ - "#14"
1291
+ - "#15"
1292
+ - "#16"
1293
+ - "#17"
1294
+ - "#18"
1295
+ - "#19"
1296
+ - "#20"
1297
+ - "#21"
1298
+ - "#22"
1299
+ - "#23"
1300
+ - "#24"
1301
+ - "#25"
1302
+ - "#26"
1303
+ - "#27"
1304
+ - "#28"
1305
+ - "#29"
1306
+ sample_n: 30
1307
+ checked_at: "2026-10-04"
1308
+ unit.state: {fill: derived, basis: "packed when pointer_level is archive, else expanded"}
1309
+ unit.usable_unit:
1310
+ fill: judgment
1311
+ basis: "every sample unpacks to one file of the declared type"
1312
+ files:
1313
+ - "#00"
1314
+ - "#01"
1315
+ - "#02"
1316
+ - "#03"
1317
+ - "#04"
1318
+ - "#05"
1319
+ - "#06"
1320
+ - "#07"
1321
+ - "#08"
1322
+ - "#09"
1323
+ - "#10"
1324
+ - "#11"
1325
+ - "#12"
1326
+ - "#13"
1327
+ - "#14"
1328
+ - "#15"
1329
+ - "#16"
1330
+ - "#17"
1331
+ - "#18"
1332
+ - "#19"
1333
+ - "#20"
1334
+ - "#21"
1335
+ - "#22"
1336
+ - "#23"
1337
+ - "#24"
1338
+ - "#25"
1339
+ - "#26"
1340
+ - "#27"
1341
+ - "#28"
1342
+ - "#29"
1343
+ sample_n: 30
1344
+ checked_at: "2026-10-04"
cards/conda-forge-packages.yaml ADDED
@@ -0,0 +1,260 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: conda-forge-packages
8
+ parent: null
9
+ title: conda-forge
10
+ description: "Package files (.conda and .tar.bz2) of every package in the conda-forge channel on conda.anaconda.org, one pointer per file. conda-forge is the community-maintained conda channel that ships builds of scientific computing and machine learning software (including CUDA, PyTorch, TensorFlow and JAX builds) for each platform. File lists, sizes and hashes come from the repodata.json of each platform subdirectory; the list of subdirectories comes from channeldata.json. Packages can contain malicious code; treat files as data and run nothing outside a sandbox."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-02"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per package (license field of repodata, mostly SPDX expressions)"
20
+ licence_url: "https://conda-forge.org"
21
+ licence_statement: "conda-forge packages are released by their upstream projects under their own open source licenses; the channel adds no license of its own."
22
+ per_collection: null
23
+ training_ok: conditional
24
+ redistribution_ok: conditional
25
+ attribution: null
26
+ per_file: "license_override.notes gives the license field of the package; license_override.spdx gives the SPDX expression where one is recognised."
27
+ licence_tiers: {unknown/unknown: 192391, yes/conditional: 728770, yes/yes: 2448084}
28
+ spdx_top:
29
+ "(none)": 192391
30
+ AGPL-3.0-only: 6591
31
+ AGPL-3.0-or-later: 6648
32
+ Apache-2.0: 823502
33
+ "Apache-2.0 AND MIT": 7216
34
+ "Apache-2.0 WITH LLVM-exception": 30841
35
+ BSD-2-Clause: 101566
36
+ BSD-3-Clause: 561801
37
+ BSD-3-Clause-LBNL: 10863
38
+ BSL-1.0: 12367
39
+ GPL-2.0: 7308
40
+ GPL-2.0-only: 58747
41
+ GPL-2.0-or-later: 114814
42
+ GPL-3.0: 12115
43
+ GPL-3.0-only: 88467
44
+ spdx_distinct: 452
45
+ licence_sample_check: null
46
+ unit:
47
+ pointer_level: null
48
+ pointer_count: 3369245
49
+ units: {file: 3369245}
50
+ state: null
51
+ expanded_count: null
52
+ bytes_total: 25368692087332
53
+ rows_without_size: 0
54
+ collections:
55
+ distinct: 34522
56
+ top:
57
+ arrow-cpp: 18691
58
+ awscli: 66365
59
+ awscrt: 19006
60
+ conda-forge-ci-setup: 8560
61
+ gdal: 15210
62
+ petsc4py: 11013
63
+ power-grid-model: 9099
64
+ pyarrow: 30820
65
+ pyarrow-tests: 29465
66
+ vim: 16078
67
+ sub_collections: null
68
+ batches: null
69
+ n_ext: 2
70
+ usable_unit: null
71
+ form:
72
+ by_ext: {conda: 2213897, bz2: 1155348}
73
+ bytes_p50: 621796
74
+ bytes_p95: 30087104
75
+ bytes_max: 2360652303
76
+ size_quantiles:
77
+ max: 2360652303
78
+ min: 313
79
+ p25: 111066
80
+ p50: 621796
81
+ p75: 3034638
82
+ p90: 13175123
83
+ p95: 30087104
84
+ p99: 121400564
85
+ p99.9: 502141444
86
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
87
+ by_ext_size:
88
+ - {ext: conda, rows: 2213897, bytes: 17953581135800, p50: 726830, p95: 32981484, max: 2360652303}
89
+ - {ext: bz2, rows: 1155348, bytes: 7415110951532, p50: 466391, p95: 23689619, max: 1328211938}
90
+ over_cap:
91
+ cap: 61944279
92
+ rows_over: 85579
93
+ bytes_over: 13578554784418
94
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
95
+ largest:
96
+ - pointer_id: "conda-forge-packages:libnvinfer:linux-aarch64/libnvinfer-11.1.0.106-h697f3a9_1.conda"
97
+ size_bytes: 2360652303
98
+ - pointer_id: "conda-forge-packages:libnvinfer:linux-aarch64/libnvinfer-10.16.1.11-h0ade831_0.conda"
99
+ size_bytes: 2350697800
100
+ - pointer_id: "conda-forge-packages:libnvinfer:linux-64/libnvinfer-11.1.0.106-h2a0a68e_1.conda"
101
+ size_bytes: 1888374427
102
+ - pointer_id: "conda-forge-packages:libnvinfer:linux-64/libnvinfer-10.16.1.11-h9f76b7a_0.conda"
103
+ size_bytes: 1882066521
104
+ - pointer_id: "conda-forge-packages:libnvinfer:win-64/libnvinfer-11.1.0.106-h9141228_1.conda"
105
+ size_bytes: 1800868888
106
+ date_years: null
107
+ media_type: {"(none)": 3369245}
108
+ facets:
109
+ format: {conda: 2213897, tar.bz2: 1155348}
110
+ ml_related: {"false": 3205598, "true": 163647}
111
+ subdir:
112
+ linux-64: 787027
113
+ linux-aarch64: 357732
114
+ linux-armv7l: 3
115
+ linux-ppc64le: 281769
116
+ linux-riscv64: 2320
117
+ linux-s390x: 5631
118
+ noarch: 380610
119
+ osx-64: 675076
120
+ osx-arm64: 337443
121
+ win-32: 24464
122
+ win-64: 513472
123
+ win-arm64: 3698
124
+ artifact: null
125
+ source_origin: null
126
+ parse_cost: null
127
+ parse_quirks: null
128
+ structure:
129
+ self_describing: null
130
+ integrity_builtin: null
131
+ has_reference_output: null
132
+ invariants: null
133
+ tools: null
134
+ ships:
135
+ problem_statement: {present: null}
136
+ reference_solution: {share: null}
137
+ tests: {present: null}
138
+ env_definition: {present: null}
139
+ known_answer: {present: null}
140
+ generator: {present: null}
141
+ upstream_labels: {present: null}
142
+ class: null
143
+ subject:
144
+ domain_span: null
145
+ shape_span: null
146
+ skill_family_span: null
147
+ ont_domain: null
148
+ tech_vocabulary: null
149
+ risk:
150
+ contamination: none
151
+ overlap_check:
152
+ done: true
153
+ pointers_flagged: 442
154
+ pointers_removed: 0
155
+ by_class: {build_dependency: 442}
156
+ benches: {Terminal-Bench: 442}
157
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
158
+ pii: null
159
+ pii_evidence: null
160
+ malware: null
161
+ ship_bytes: conditional
162
+ subset:
163
+ rule: "take all rows; licence, size and version are labels, not filters"
164
+ rows_in_subset: 3369245
165
+ last_built: null
166
+ access:
167
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
168
+ verify: "Compare size_bytes, sha256 and md5 (published in repodata)."
169
+ hosts: {"https://conda.anaconda.org": 3369245}
170
+ uri_example: "https://conda.anaconda.org/conda-forge/linux-64/21cmfast-3.0.2-py36h1af98f8_1.tar.bz2"
171
+ needs_auth: {"false": 3369245}
172
+ credential: null
173
+ hash_source: {source_published: 3369245}
174
+ range_supported: null
175
+ reachable:
176
+ index_flag: {unknown: 3369245}
177
+ index_check: null
178
+ our_sample: null
179
+ drift: null
180
+ sample: null
181
+ evidence:
182
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
184
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ class: {fill: unfilled, basis: "ships not judged"}
186
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ form.by_ext:
188
+ fill: auto
189
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
190
+ checked_at: "2026-10-03"
191
+ form.date_years:
192
+ fill: auto
193
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
194
+ checked_at: "2026-10-03"
195
+ form.over_cap:
196
+ fill: auto
197
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
198
+ checked_at: "2026-10-03"
199
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ form.size_quantiles:
202
+ fill: auto
203
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
204
+ checked_at: "2026-10-03"
205
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
207
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ provenance.licence_tiers:
209
+ fill: auto
210
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
211
+ checked_at: "2026-10-03"
212
+ provenance.spdx_top:
213
+ fill: auto
214
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
215
+ checked_at: "2026-10-03"
216
+ risk.contamination:
217
+ fill: auto
218
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
219
+ checked_at: "2026-10-03"
220
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
224
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
228
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ snapshot.rev:
237
+ fill: auto
238
+ basis: "last commit listed in the publish receipt of this source"
239
+ checked_at: "2026-10-03"
240
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
241
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
242
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
243
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
244
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
245
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
247
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
249
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
250
+ subset:
251
+ fill: auto
252
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
253
+ checked_at: "2026-10-03"
254
+ unit.pointer_count:
255
+ fill: auto
256
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
257
+ checked_at: "2026-10-03"
258
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
260
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/cran.yaml ADDED
@@ -0,0 +1,248 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: cran
8
+ parent: null
9
+ title: CRAN
10
+ description: "Source packages of all current R packages on CRAN (src/contrib). The list and md5 come from the PACKAGES index; sizes come from HEAD requests."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: "542105e2239faa985b30f22d79dad7ff25fdc6eb", at: "2026-10-03", rev_candidates: 2}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "Per package (GPL, MIT, Apache and others)"
17
+ licence_url: "https://cran.r-project.org"
18
+ licence_statement: "CRAN accepts only packages under open source licenses; each package declares its own."
19
+ per_collection: null
20
+ training_ok: "yes"
21
+ redistribution_ok: conditional
22
+ attribution: null
23
+ per_file: "license_override.notes gives the License field from PACKAGES; license_override.spdx gives the SPDX id where one is recognised."
24
+ licence_tiers: {yes/conditional: 16994, yes/yes: 8305}
25
+ spdx_top:
26
+ "(none)": 659
27
+ AGPL-3.0-only: 152
28
+ Apache-2.0: 594
29
+ Artistic-2.0: 99
30
+ BSD-2-Clause: 133
31
+ BSD-3-Clause: 168
32
+ BSL-1.0: 8
33
+ CC0-1.0: 253
34
+ CECILL-2.1: 2
35
+ CPL-1.0: 1
36
+ EPL-1.0: 1
37
+ EUPL-1.1: 1
38
+ EUPL-1.2: 7
39
+ GPL-2.0-only: 2585
40
+ GPL-2.0-or-later: 4743
41
+ spdx_distinct: 25
42
+ licence_sample_check: null
43
+ unit:
44
+ pointer_level: null
45
+ pointer_count: 25299
46
+ units: {file: 25299}
47
+ state: null
48
+ expanded_count: null
49
+ bytes_total: 18898338630
50
+ rows_without_size: 0
51
+ collections:
52
+ distinct: 25284
53
+ top:
54
+ KernSmooth: 2
55
+ MASS: 2
56
+ Matrix: 2
57
+ boot: 2
58
+ class: 2
59
+ cluster: 2
60
+ codetools: 2
61
+ foreign: 2
62
+ lattice: 2
63
+ mgcv: 2
64
+ sub_collections: null
65
+ batches: null
66
+ n_ext: 1
67
+ usable_unit: null
68
+ form:
69
+ by_ext: {gz: 25299}
70
+ bytes_p50: 165496
71
+ bytes_p95: 3578835
72
+ bytes_max: 21896924
73
+ size_quantiles:
74
+ max: 21896924
75
+ min: 1090
76
+ p25: 28366
77
+ p50: 165496
78
+ p75: 855652
79
+ p90: 2416000
80
+ p95: 3578835
81
+ p99: 5429666
82
+ p99.9: 11607681
83
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
84
+ by_ext_size:
85
+ - {ext: gz, rows: 25299, bytes: 18898338630, p50: 165496, p95: 3578835, max: 21896924}
86
+ over_cap:
87
+ cap: 61944279
88
+ rows_over: 0
89
+ bytes_over: 0
90
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
91
+ largest:
92
+ - {pointer_id: "cran:tantivyr:tantivyr_0.1.1.tar.gz", size_bytes: 21896924}
93
+ - {pointer_id: "cran:arcgisplaces:arcgisplaces_0.1.2.tar.gz", size_bytes: 21589598}
94
+ - {pointer_id: "cran:rcdklibs:rcdklibs_2.9.tar.gz", size_bytes: 19457497}
95
+ - {pointer_id: "cran:pdfsigner:pdfsigner_0.3.0.tar.gz", size_bytes: 18465357}
96
+ - {pointer_id: "cran:fastrmodels:fastrmodels_2.1.0.tar.gz", size_bytes: 16883666}
97
+ date_years: null
98
+ media_type: {"(none)": 25299}
99
+ facets:
100
+ needs_compilation: {"no": 19767, "yes": 5532}
101
+ artifact: null
102
+ source_origin: null
103
+ parse_cost: null
104
+ parse_quirks: null
105
+ structure:
106
+ self_describing: null
107
+ integrity_builtin: null
108
+ has_reference_output: null
109
+ invariants: null
110
+ tools: null
111
+ ships:
112
+ problem_statement: {present: null}
113
+ reference_solution: {share: null}
114
+ tests: {present: null}
115
+ env_definition: {present: null}
116
+ known_answer: {present: null}
117
+ generator: {present: null}
118
+ upstream_labels: {present: null}
119
+ class: null
120
+ subject:
121
+ domain_span: null
122
+ shape_span: null
123
+ skill_family_span: null
124
+ ont_domain: null
125
+ tech_vocabulary: null
126
+ risk:
127
+ contamination: none
128
+ overlap_check:
129
+ done: true
130
+ pointers_flagged: 0
131
+ pointers_removed: 0
132
+ by_class: {}
133
+ benches: {}
134
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
135
+ pii: null
136
+ pii_evidence: null
137
+ malware: null
138
+ ship_bytes: conditional
139
+ subset:
140
+ rule: "take all rows; licence, size and version are labels, not filters"
141
+ rows_in_subset: 25299
142
+ last_built:
143
+ date: "2026-10-02"
144
+ rule: "older rule that filtered by licence tier"
145
+ rows: 25299
146
+ rows_in_subset: 25299
147
+ excluded_by_licence: 0
148
+ duplicates: 0
149
+ access:
150
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
151
+ verify: "Compare size_bytes and md5 (MD5sum in PACKAGES). If the current address returns 404, try the Archive address in alt_uris."
152
+ hosts: {"https://cloud.r-project.org": 25299}
153
+ uri_example: "https://cloud.r-project.org/src/contrib/AATtools_0.0.3.tar.gz"
154
+ needs_auth: {"false": 25299}
155
+ credential: null
156
+ hash_source: {source_published: 25299}
157
+ range_supported: null
158
+ reachable:
159
+ index_flag: {unknown: 25299}
160
+ index_check:
161
+ checked_at: "2026-10-02"
162
+ sample: 30
163
+ checked: 25284
164
+ ok: 25284
165
+ by_status: {ok: 25284}
166
+ our_sample: null
167
+ drift: null
168
+ sample: null
169
+ evidence:
170
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
171
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
172
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
173
+ class: {fill: unfilled, basis: "ships not judged"}
174
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
175
+ form.by_ext:
176
+ fill: auto
177
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
178
+ checked_at: "2026-10-03"
179
+ form.date_years:
180
+ fill: auto
181
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
182
+ checked_at: "2026-10-03"
183
+ form.over_cap:
184
+ fill: auto
185
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
186
+ checked_at: "2026-10-03"
187
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
188
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
189
+ form.size_quantiles:
190
+ fill: auto
191
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
192
+ checked_at: "2026-10-03"
193
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
194
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
195
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
196
+ provenance.licence_tiers:
197
+ fill: auto
198
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
199
+ checked_at: "2026-10-03"
200
+ provenance.spdx_top:
201
+ fill: auto
202
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
203
+ checked_at: "2026-10-03"
204
+ risk.contamination:
205
+ fill: auto
206
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
207
+ checked_at: "2026-10-03"
208
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
212
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
213
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
214
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ snapshot.rev:
225
+ fill: auto
226
+ basis: "last commit listed in the publish receipt of this source"
227
+ checked_at: "2026-10-03"
228
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
237
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
238
+ subset:
239
+ fill: auto
240
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
241
+ checked_at: "2026-10-03"
242
+ unit.pointer_count:
243
+ fill: auto
244
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
245
+ checked_at: "2026-10-03"
246
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
247
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
248
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/crates-io.yaml ADDED
@@ -0,0 +1,489 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: crates-io
8
+ parent: null
9
+ title: crates.io
10
+ description: "The .crate file of every version of every crate on crates.io, from the official daily database dump (2026-10-01T02:00:25.505616213Z)."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-01"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ dump: {crates_in_dump: 343129, dump_timestamp: "2026-10-01T02:00:25.505616213Z"}
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: "Per version (license field of Cargo.toml, SPDX expression)"
21
+ licence_url: "https://crates.io"
22
+ licence_statement: "Each crate version declares its own license."
23
+ per_collection: null
24
+ training_ok: conditional
25
+ redistribution_ok: conditional
26
+ attribution: null
27
+ per_file: "license_override.notes gives the license field; license_override.spdx gives the SPDX expression where one is recognised."
28
+ licence_tiers: {unknown/unknown: 240087, yes/conditional: 285196, yes/yes: 2726862}
29
+ spdx_top:
30
+ "(none)": 240087
31
+ "0BSD": 10515
32
+ AGPL-3.0: 13987
33
+ AGPL-3.0-only: 22131
34
+ AGPL-3.0-or-later: 26988
35
+ Apache-2.0: 633672
36
+ "Apache-2.0 OR MIT": 68845
37
+ "Apache-2.0 WITH LLVM-exception": 16333
38
+ "Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT": 3880
39
+ BSD-2-Clause: 12667
40
+ BSD-3-Clause: 45707
41
+ CC0-1.0: 10979
42
+ EPL-1.0: 6537
43
+ EUPL-1.2: 8116
44
+ GPL-2.0: 3392
45
+ spdx_distinct: 669
46
+ licence_sample_check: {sample_n: 10, declared: 10, unknown_or_missing: 0}
47
+ unit:
48
+ pointer_level: archive
49
+ pointer_count: 3252145
50
+ units: {file: 3252145}
51
+ state: packed
52
+ expanded_count: null
53
+ bytes_total: 663497168445
54
+ rows_without_size: 0
55
+ collections:
56
+ distinct: 343129
57
+ top:
58
+ icydb: 1410
59
+ icydb-core: 1423
60
+ icydb-schema: 1424
61
+ spider: 2227
62
+ spider_cli: 2153
63
+ spider_utils: 1479
64
+ spider_worker: 1988
65
+ swc: 1452
66
+ swc_core: 1492
67
+ swc_ecma_minifier: 1366
68
+ sub_collections: null
69
+ batches: null
70
+ n_ext: 1
71
+ usable_unit: "one .crate file of one version of one crate; successive versions of a crate overlap heavily"
72
+ form:
73
+ by_ext: {crate: 3252145}
74
+ bytes_p50: 29485
75
+ bytes_p95: 758413
76
+ bytes_max: 42206040
77
+ size_quantiles:
78
+ max: 42206040
79
+ min: 53
80
+ p25: 11502
81
+ p50: 29485
82
+ p75: 80632
83
+ p90: 300525
84
+ p95: 758413
85
+ p99: 4190204
86
+ p99.9: 9483693
87
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
88
+ by_ext_size:
89
+ - {ext: crate, rows: 3252145, bytes: 663497168445, p50: 29485, p95: 758413, max: 42206040}
90
+ over_cap:
91
+ cap: 61944279
92
+ rows_over: 0
93
+ bytes_over: 0
94
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
95
+ largest:
96
+ - {pointer_id: "crates-io:pv_leopard:pv_leopard-2.0.2.crate", size_bytes: 42206040}
97
+ - {pointer_id: "crates-io:pv_leopard:pv_leopard-2.0.3.crate", size_bytes: 41389212}
98
+ - {pointer_id: "crates-io:lindera-ko-dic:lindera-ko-dic-0.27.1.crate", size_bytes: 41068953}
99
+ - {pointer_id: "crates-io:lindera-ko-dic:lindera-ko-dic-0.26.0.crate", size_bytes: 41068951}
100
+ - {pointer_id: "crates-io:lindera-ko-dic:lindera-ko-dic-0.27.2.crate", size_bytes: 41068947}
101
+ date_years:
102
+ key: created_at
103
+ rows_with_date: 3252145
104
+ counts:
105
+ "2014": 1753
106
+ "2015": 17558
107
+ "2016": 20341
108
+ "2017": 33705
109
+ "2018": 47234
110
+ "2019": 71535
111
+ "2020": 117624
112
+ "2021": 150262
113
+ "2022": 212897
114
+ "2023": 300135
115
+ "2024": 382383
116
+ "2025": 506547
117
+ "2026": 1390171
118
+ media_type: {"(none)": 3252145}
119
+ facets:
120
+ yanked: {"false": 3133151, "true": 118994}
121
+ artifact: [source-code, configuration]
122
+ source_origin: [downloaded-source-archive]
123
+ parse_cost: needs_decompress
124
+ parse_quirks:
125
+ - "Cargo.toml is the registry-normalized copy; the author's original is Cargo.toml.orig"
126
+ - "a crate cut from a workspace may refer to sibling crates only by registry version"
127
+ structure:
128
+ self_describing: true
129
+ integrity_builtin: [gzip_crc, cargo_lock_checksums]
130
+ has_reference_output: false
131
+ invariants: [sha256_matches_index, cargo_toml_name_version_match_filename]
132
+ tools: [tar, cargo]
133
+ ships:
134
+ problem_statement: {present: false}
135
+ reference_solution:
136
+ share: {k: 0, "n": 10}
137
+ tests:
138
+ present: true
139
+ coverage: "3 of 10 samples carry test files (tests/ directory in #01, #06, #08; #08 also has src/**/tests.rs modules); inline #[cfg(test)] blocks inside src files are not visible in the digests, so this is a lower bound"
140
+ runs: null
141
+ env_definition: {present: true, kind: dependency_manifest, runs: null}
142
+ known_answer: {present: false}
143
+ generator: {present: false}
144
+ upstream_labels:
145
+ present: true
146
+ fields: [keywords, categories, license]
147
+ class: manifest_not_run
148
+ subject:
149
+ domain_span: [D1, D2.1, D2.3]
150
+ shape_span: [S1_code_to_tests, S2_env_build_config]
151
+ skill_family_span: [skill.software, skill.systems]
152
+ ont_domain: [software-systems, ml-ai]
153
+ tech_vocabulary: [rust, cargo, Cargo.toml, Cargo.lock, crate, "edition 2021", "edition 2024", build.rs]
154
+ risk:
155
+ contamination: public_benchmark
156
+ overlap_check:
157
+ done: true
158
+ pointers_flagged: 2
159
+ pointers_removed: 1
160
+ by_class: {build_dependency: 1, derivative_container: 1}
161
+ benches: {"FrontierSWE v2": 1, swebench: 1}
162
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
163
+ pii: present
164
+ pii_evidence: {sample_n: 10, files_with_email: 7, note: "authors field in Cargo.toml; public attribution"}
165
+ malware: null
166
+ ship_bytes: conditional
167
+ subset:
168
+ rule: "take all rows; licence, size and version are labels, not filters"
169
+ rows_in_subset: 3252145
170
+ last_built: null
171
+ access:
172
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
173
+ verify: "Compare size_bytes and sha256 (published in the dump)."
174
+ hosts: {"https://static.crates.io": 3252145}
175
+ uri_example: "https://static.crates.io/crates/A-Mazed/A-Mazed-0.1.0.crate"
176
+ needs_auth: {"false": 3252145}
177
+ credential: none
178
+ hash_source: {source_published: 3252145}
179
+ range_supported: null
180
+ reachable:
181
+ index_flag: {unknown: 3252145}
182
+ index_check: null
183
+ our_sample: {checked_at: "2026-10-04", n_tried: 10, n_ok: 10, hash_checked: 10}
184
+ drift: "published crates are immutable (yank hides but does not delete); new versions arrive daily"
185
+ sample:
186
+ method: "strata are file type times the size quartile of this source (q1 to q4 by its own p25, p50, p75; unk for rows without a size); half of the samples go one per stratum to the largest strata, the rest in proportion to stratum size (largest remainder); within a stratum the rows with the smallest sha256(seed:pointer_id) are taken, so the draw is reproducible from the index"
187
+ seed: 20261004
188
+ "n": 10
189
+ n_drawn: 10
190
+ n_read: 10
191
+ unread: []
192
+ rule_n: "10 when one file type holds at least 95 percent of rows, else 30"
193
+ size_quartiles: [11502, 29485, 80632]
194
+ size_filter: "only files of at most 15000000 bytes are candidates"
195
+ strata_candidates: {"crate|q1": 813082, "crate|q3": 813023, "crate|q2": 813007, "crate|q4": 812596}
196
+ strata_total: 4
197
+ strata_taken: {"crate|q1": 813082, "crate|q2": 813007, "crate|q3": 813023, "crate|q4": 812596}
198
+ fetched_at: "2026-10-04"
199
+ replaced: 0
200
+ kept: false
201
+ kept_note: "sample bytes were deleted after reading; refetch by pointer_id and compare sha256"
202
+ files:
203
+ - "no": "#00"
204
+ pointer_id: "crates-io:embedded-savegame:embedded-savegame-0.1.0.crate"
205
+ stratum: "crate|q1"
206
+ size_bytes: 10772
207
+ sha256: "760082c11316e524c5da1d8358464e91b822cb4e7f1bd29aa4945f04306d84d7"
208
+ - "no": "#01"
209
+ pointer_id: "crates-io:restruct:restruct-0.1.0.crate"
210
+ stratum: "crate|q1"
211
+ size_bytes: 9436
212
+ sha256: d91fafdd890db1341e28784db6e09efe1b04abb2b8a12d819ea626b182b240e4
213
+ - "no": "#02"
214
+ pointer_id: "crates-io:xtagger:xtagger-1.1.0.crate"
215
+ stratum: "crate|q1"
216
+ size_bytes: 11144
217
+ sha256: "544bd86dfc4f60b128a1b2ab7eabb6fd9b0223866cb2243109766620b845c147"
218
+ - "no": "#03"
219
+ pointer_id: "crates-io:naia-bevy-server:naia-bevy-server-0.24.0.crate"
220
+ stratum: "crate|q2"
221
+ size_bytes: 18486
222
+ sha256: b5e5fbd574154cc5f02989759f40bd064ae2c8cfc27cf9df2f659b97e7bd7a4e
223
+ - "no": "#04"
224
+ pointer_id: "crates-io:chia-protocol:chia-protocol-0.2.14.crate"
225
+ stratum: "crate|q2"
226
+ size_bytes: 15959
227
+ sha256: c067b2023d59ccd9c9eeeebbec8caabe1010cb97b542ba87451a11299647f8e8
228
+ - "no": "#05"
229
+ pointer_id: "crates-io:xlsynth-autocov:xlsynth-autocov-0.56.0.crate"
230
+ stratum: "crate|q3"
231
+ size_bytes: 49588
232
+ sha256: fd1ab552cfed7e1148a82d855ea90d390e4daa4f6bbd4f23db4af12326f96ae2
233
+ - "no": "#06"
234
+ pointer_id: "crates-io:typeweld-ls:typeweld-ls-0.0.4.crate"
235
+ stratum: "crate|q3"
236
+ size_bytes: 54307
237
+ sha256: ebfb584a5983c9d1f811caaf22cf3f4a01982a3002ea071f63bd1c78742eb705
238
+ - "no": "#07"
239
+ pointer_id: "crates-io:mockforge-workspace:mockforge-workspace-0.3.178.crate"
240
+ stratum: "crate|q3"
241
+ size_bytes: 52823
242
+ sha256: "6354d7290c0ae7edec6b52fde2bf65af6434a10d6b07976757fd040b506d6cdd"
243
+ - "no": "#08"
244
+ pointer_id: "crates-io:lenso-engine-authoring:lenso-engine-authoring-0.2.3.crate"
245
+ stratum: "crate|q4"
246
+ size_bytes: 131988
247
+ sha256: "7c67bdb619120ff386fe8a83f5cc8454f921fb3944038400a47dd7b3f4565dee"
248
+ - "no": "#09"
249
+ pointer_id: "crates-io:adk-server:adk-server-0.1.7.crate"
250
+ stratum: "crate|q4"
251
+ size_bytes: 1338250
252
+ sha256: dd82941065b0dc331b510e09bf080a2d09c924b01ac3b3227bc233d827082144
253
+ evidence:
254
+ access.credential:
255
+ fill: judgment
256
+ basis: "all 10 fetched anonymously"
257
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
258
+ sample_n: 10
259
+ checked_at: "2026-10-04"
260
+ access.drift:
261
+ fill: judgment
262
+ basis: "all 10 sha256 matched the index"
263
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
264
+ sample_n: 10
265
+ checked_at: "2026-10-04"
266
+ access.range_supported: {fill: unfilled, basis: "no sample evidence given for this field"}
267
+ class:
268
+ fill: derived
269
+ basis: "rule over ships.tests, env_definition, known_answer, generator and structure; manifest_not_run when tests and a dependency manifest are present but the build was not run"
270
+ form.artifact:
271
+ fill: judgment
272
+ basis: "Rust source files (.rs) are most members in all 10; Cargo.toml and Cargo.lock are build configuration"
273
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
274
+ sample_n: 10
275
+ checked_at: "2026-10-04"
276
+ form.by_ext:
277
+ fill: auto
278
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
279
+ checked_at: "2026-10-03"
280
+ form.date_years:
281
+ fill: auto
282
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
283
+ checked_at: "2026-10-03"
284
+ form.over_cap:
285
+ fill: auto
286
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
287
+ checked_at: "2026-10-03"
288
+ form.parse_cost:
289
+ fill: judgment
290
+ basis: "gzip tar, must be unpacked before reading"
291
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
292
+ sample_n: 10
293
+ checked_at: "2026-10-04"
294
+ form.parse_quirks:
295
+ fill: judgment
296
+ basis: "Cargo.toml headers say AUTOMATICALLY GENERATED BY CARGO and point to Cargo.toml.orig (#08); .orig present in #01, #03 to #08"
297
+ files: ["#01", "#03", "#08"]
298
+ sample_n: 10
299
+ checked_at: "2026-10-04"
300
+ form.size_quantiles:
301
+ fill: auto
302
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
303
+ checked_at: "2026-10-03"
304
+ form.source_origin:
305
+ fill: judgment
306
+ basis: "crate archives fetched by URL from the static crates host; .cargo_vcs_info.json names a git commit but no git history ships"
307
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
308
+ sample_n: 10
309
+ checked_at: "2026-10-04"
310
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
311
+ provenance.licence_sample_check:
312
+ fill: judgment
313
+ basis: "license field in Cargo.toml in all 10 (MIT, Apache-2.0, or MIT OR Apache-2.0)"
314
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
315
+ sample_n: 10
316
+ checked_at: "2026-10-04"
317
+ provenance.licence_tiers:
318
+ fill: auto
319
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
320
+ checked_at: "2026-10-03"
321
+ provenance.spdx_top:
322
+ fill: auto
323
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
324
+ checked_at: "2026-10-03"
325
+ risk.contamination:
326
+ fill: auto
327
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
328
+ checked_at: "2026-10-03"
329
+ risk.malware: {fill: unfilled, basis: "no sample evidence given for this field"}
330
+ risk.pii:
331
+ fill: judgment
332
+ basis: "author email addresses in Cargo.toml authors in 7 of 10 samples; public attribution; re-checked under the WP-T-44c definition (a user name on a platform counts as present)"
333
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#09"]
334
+ sample_n: 10
335
+ checked_at: "2026-10-04"
336
+ risk.pii_evidence:
337
+ fill: judgment
338
+ basis: "email regex over the digests"
339
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
340
+ sample_n: 10
341
+ checked_at: "2026-10-04"
342
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
343
+ ships.env_definition.kind:
344
+ fill: judgment
345
+ basis: "Cargo.toml plus Cargo.lock pin the dependency graph; no container or toolchain image ships"
346
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
347
+ sample_n: 10
348
+ checked_at: "2026-10-04"
349
+ ships.env_definition.present:
350
+ fill: judgment
351
+ basis: "Cargo.toml in all 10, Cargo.lock in 8 of 10, build.rs in #08"
352
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
353
+ sample_n: 10
354
+ checked_at: "2026-10-04"
355
+ ships.env_definition.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
356
+ ships.generator.present:
357
+ fill: judgment
358
+ basis: "nothing produces (input, answer) pairs"
359
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
360
+ sample_n: 10
361
+ checked_at: "2026-10-04"
362
+ ships.known_answer.present:
363
+ fill: judgment
364
+ basis: "no expected value file"
365
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
366
+ sample_n: 10
367
+ checked_at: "2026-10-04"
368
+ ships.problem_statement.present:
369
+ fill: judgment
370
+ basis: "no task text; README and description only describe the crate"
371
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
372
+ sample_n: 10
373
+ checked_at: "2026-10-04"
374
+ ships.reference_solution.share:
375
+ fill: judgment
376
+ basis: "0 of 10 sample files read: no sample file carries a fix, patch, solved answer or worked solution"
377
+ files: []
378
+ sample_n: 10
379
+ checked_at: "2026-10-04"
380
+ ships.tests.coverage:
381
+ fill: judgment
382
+ basis: "member lists and test-like path counts in the digests"
383
+ files: ["#01", "#06", "#08"]
384
+ sample_n: 10
385
+ checked_at: "2026-10-04"
386
+ ships.tests.present:
387
+ fill: judgment
388
+ basis: "test files or test modules in 3 of 10 samples"
389
+ files: ["#01", "#06", "#08"]
390
+ sample_n: 10
391
+ checked_at: "2026-10-04"
392
+ ships.tests.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
393
+ ships.upstream_labels.fields:
394
+ fill: judgment
395
+ basis: "fields in the normalized Cargo.toml"
396
+ files: ["#02", "#07"]
397
+ sample_n: 10
398
+ checked_at: "2026-10-04"
399
+ ships.upstream_labels.present:
400
+ fill: judgment
401
+ basis: "keywords or categories arrays in Cargo.toml in 4 of 10 (#02, #06, #07, #09; #07 is in development-tools::testing)"
402
+ files: ["#02", "#06", "#07", "#09"]
403
+ sample_n: 10
404
+ checked_at: "2026-10-04"
405
+ snapshot.rev:
406
+ fill: auto
407
+ basis: "last commit listed in the publish receipt of this source"
408
+ checked_at: "2026-10-03"
409
+ structure.has_reference_output:
410
+ fill: judgment
411
+ basis: "no expected output file in any sample"
412
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
413
+ sample_n: 10
414
+ checked_at: "2026-10-04"
415
+ structure.integrity_builtin:
416
+ fill: judgment
417
+ basis: "gzip CRC; Cargo.lock lists a checksum per dependency (8 of 10 ship Cargo.lock: #00, #02, #03, #05 to #09)"
418
+ files: ["#00", "#02", "#03", "#05", "#06", "#07", "#08", "#09"]
419
+ sample_n: 10
420
+ checked_at: "2026-10-04"
421
+ structure.invariants:
422
+ fill: judgment
423
+ basis: "all 10 fetched files matched the sha256 in the index"
424
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
425
+ sample_n: 10
426
+ checked_at: "2026-10-04"
427
+ structure.self_describing:
428
+ fill: judgment
429
+ basis: "Cargo.toml with name, version, edition, license and dependencies in all 10"
430
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
431
+ sample_n: 10
432
+ checked_at: "2026-10-04"
433
+ structure.tools:
434
+ fill: judgment
435
+ basis: "unpack with tar; build and test with cargo"
436
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
437
+ sample_n: 10
438
+ checked_at: "2026-10-04"
439
+ subject.domain_span:
440
+ fill: judgment
441
+ basis: "libraries and servers (#03 game networking, #04 protocol types, #09 HTTP agent server); language-server tooling (#06); XLS IR coverage tooling for hardware synthesis (#05); embedded storage with wear levelling (#00); build scripts and dependency locks in most samples; trimmed to at most three main values (WP-T-44c): kept the values most samples show, dropped a parent code when its subcode is kept"
442
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
443
+ sample_n: 10
444
+ checked_at: "2026-10-04"
445
+ subject.ont_domain:
446
+ fill: judgment
447
+ basis: "all 10 are software libraries or tools; #09 is an agent development kit server"
448
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
449
+ sample_n: 10
450
+ checked_at: "2026-10-04"
451
+ subject.shape_span:
452
+ fill: judgment
453
+ basis: "crates with tests can seed fix-to-pass tasks (#01, #06, #08); Cargo builds can seed build tasks (all)"
454
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
455
+ sample_n: 10
456
+ checked_at: "2026-10-04"
457
+ subject.skill_family_span:
458
+ fill: judgment
459
+ basis: "Rust programming; building with cargo"
460
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
461
+ sample_n: 10
462
+ checked_at: "2026-10-04"
463
+ subject.tech_vocabulary:
464
+ fill: judgment
465
+ basis: "terms in member lists and Cargo.toml (editions 2018, 2021, 2024 seen)"
466
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
467
+ sample_n: 10
468
+ checked_at: "2026-10-04"
469
+ subset:
470
+ fill: auto
471
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
472
+ checked_at: "2026-10-03"
473
+ unit.pointer_count:
474
+ fill: auto
475
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
476
+ checked_at: "2026-10-03"
477
+ unit.pointer_level:
478
+ fill: judgment
479
+ basis: "every sample is one .crate file, a gzip tar of one crate version's source tree"
480
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
481
+ sample_n: 10
482
+ checked_at: "2026-10-04"
483
+ unit.state: {fill: derived, basis: "packed when pointer_level is archive, else expanded"}
484
+ unit.usable_unit:
485
+ fill: judgment
486
+ basis: "10 samples are 10 different crates at one version each"
487
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
488
+ sample_n: 10
489
+ checked_at: "2026-10-04"
cards/ctftime.yaml ADDED
@@ -0,0 +1,217 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: ctftime
8
+ parent: null
9
+ title: "CTFtime events, tasks and writeups"
10
+ description: "Catalog of CTF (capture the flag) security competitions from CTFtime as of 2026-10-02: events (CTF competitions, from the CTFtime API); tasks (challenges, from the public task list); writeups (from the public writeup list, each writeup page fetched once). One pointer per event, task and writeup. Event pointers hash the single-event API response as returned; task and writeup pointers hash the stable part of the page; writeup pointers record the original writeup URL and its host."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "No content license stated"
20
+ licence_url: "https://ctftime.org/api/"
21
+ licence_statement: "CTFtime states no content license and has no terms page. Site footer: \"All tasks and writeups are copyrighted by their respective authors.\" API page: \"This API is provided for data analysis and mobile applications only. You can not use this API to run CTFtime clones\"; \"most of the CTFtime data is moderated by humans, please, respect their time.\" robots.txt for all agents: \"Content-signal: search=yes, ai-train=no, use=reference\". Not for training."
22
+ per_collection: null
23
+ training_ok: "no"
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {no/unknown: 75162}
28
+ spdx_top: {"(none)": 75162}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 75162
34
+ units: {"API response": 2930, file: 72232}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 488918598
38
+ rows_without_size: 33023
39
+ collections:
40
+ distinct: 3
41
+ top: {events: 2930, tasks: 33023, writeups: 39209}
42
+ sub_collections: null
43
+ batches: null
44
+ n_ext: 2
45
+ usable_unit: null
46
+ form:
47
+ by_ext: {html: 72232, json: 2930}
48
+ bytes_p50: 1517
49
+ bytes_p95: 31824
50
+ bytes_max: 1408215
51
+ size_quantiles:
52
+ max: 1408215
53
+ min: 510
54
+ p25: 1114
55
+ p50: 1517
56
+ p75: 5603
57
+ p90: 16631
58
+ p95: 31824
59
+ p99: 313657
60
+ p99.9: 379199
61
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
62
+ by_ext_size:
63
+ - {ext: html, rows: 72232, bytes: 485167373, p50: 1593, p95: 34227, max: 1408215}
64
+ - {ext: json, rows: 2930, bytes: 3751225, p50: 1116, p95: 2429, max: 11379}
65
+ over_cap:
66
+ cap: 61944279
67
+ rows_over: 0
68
+ bytes_over: 0
69
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
70
+ largest:
71
+ - {pointer_id: "ctftime:writeups:13007", size_bytes: 1408215}
72
+ - {pointer_id: "ctftime:writeups:33302", size_bytes: 1232146}
73
+ - {pointer_id: "ctftime:writeups:31921", size_bytes: 1059828}
74
+ - {pointer_id: "ctftime:writeups:30588", size_bytes: 933347}
75
+ - {pointer_id: "ctftime:writeups:23216", size_bytes: 813339}
76
+ date_years: null
77
+ media_type: {application/json: 2930, text/html: 72232}
78
+ facets:
79
+ format: {Attack-Defense: 251, "Hack quest": 54, Jeopardy: 2624}
80
+ onsite: {"false": 2299, "true": 631}
81
+ artifact: null
82
+ source_origin: null
83
+ parse_cost: null
84
+ parse_quirks: null
85
+ structure:
86
+ self_describing: null
87
+ integrity_builtin: null
88
+ has_reference_output: null
89
+ invariants: null
90
+ tools: null
91
+ ships:
92
+ problem_statement: {present: null}
93
+ reference_solution: {share: null}
94
+ tests: {present: null}
95
+ env_definition: {present: null}
96
+ known_answer: {present: null}
97
+ generator: {present: null}
98
+ upstream_labels: {present: null}
99
+ class: null
100
+ subject:
101
+ domain_span: null
102
+ shape_span: null
103
+ skill_family_span: null
104
+ ont_domain: null
105
+ tech_vocabulary: null
106
+ risk:
107
+ contamination: none
108
+ overlap_check:
109
+ done: true
110
+ pointers_flagged: 0
111
+ pointers_removed: 0
112
+ by_class: {}
113
+ benches: {}
114
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
115
+ pii: null
116
+ pii_evidence: null
117
+ malware: null
118
+ ship_bytes: unknown
119
+ subset:
120
+ rule: "take all rows; licence, size and version are labels, not filters"
121
+ rows_in_subset: 75162
122
+ last_built: null
123
+ access:
124
+ method: "GET uri. Event uris are the single-event API (JSON); task and writeup uris are HTML pages. robots.txt asks for \"Crawl-delay: 10\" (seconds between requests)."
125
+ verify: "Events: compare size and sha256 of the response body exactly as returned, without parsing and reserializing it (hash_scope api_body, api_body_field $). Tasks and writeups (hash_scope html_region, source_meta.html_region ctftime_page): extract the stable part of the page as described in retrieval.md and compare its size and sha256 when given. Tasks, and writeups without a hash, publish no size or hash: compute them over the extracted part on retrieval."
126
+ hosts: {"https://ctftime.org": 75162}
127
+ uri_example: "https://ctftime.org/api/v1/events/1/"
128
+ needs_auth: {"false": 75162}
129
+ credential: null
130
+ hash_source: {computed: 42139, computed_on_fetch: 33023}
131
+ range_supported: null
132
+ reachable:
133
+ index_flag: {unknown: 75162}
134
+ index_check: null
135
+ our_sample: null
136
+ drift: null
137
+ sample: null
138
+ evidence:
139
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
140
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
141
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
142
+ class: {fill: unfilled, basis: "ships not judged"}
143
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
144
+ form.by_ext:
145
+ fill: auto
146
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
147
+ checked_at: "2026-10-03"
148
+ form.date_years:
149
+ fill: auto
150
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
151
+ checked_at: "2026-10-03"
152
+ form.over_cap:
153
+ fill: auto
154
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
155
+ checked_at: "2026-10-03"
156
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
157
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
158
+ form.size_quantiles:
159
+ fill: auto
160
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
161
+ checked_at: "2026-10-03"
162
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
163
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
164
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
165
+ provenance.licence_tiers:
166
+ fill: auto
167
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
168
+ checked_at: "2026-10-03"
169
+ provenance.spdx_top:
170
+ fill: auto
171
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
172
+ checked_at: "2026-10-03"
173
+ risk.contamination:
174
+ fill: auto
175
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
176
+ checked_at: "2026-10-03"
177
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
178
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
179
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
180
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
181
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
184
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
186
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
188
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
189
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
190
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
191
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
192
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
193
+ snapshot.rev:
194
+ fill: auto
195
+ basis: "last commit listed in the publish receipt of this source"
196
+ checked_at: "2026-10-03"
197
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
198
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
199
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ subset:
208
+ fill: auto
209
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
210
+ checked_at: "2026-10-03"
211
+ unit.pointer_count:
212
+ fill: auto
213
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
214
+ checked_at: "2026-10-03"
215
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
217
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/datacite-dois.yaml ADDED
@@ -0,0 +1,308 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: datacite-dois
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers: {conditional/conditional: 3162, conditional/no: 79, unknown/unknown: 9497774, yes/yes: 2490}
29
+ spdx_top:
30
+ "(none)": 9497774
31
+ BSD-3-Clause: 3
32
+ CC-BY-4.0: 2269
33
+ CC-BY-NC-4.0: 96
34
+ CC-BY-NC-ND-4.0: 73
35
+ CC-BY-NC-SA-4.0: 3011
36
+ CC-BY-ND-4.0: 6
37
+ CC-BY-SA-4.0: 55
38
+ CC0-1.0: 216
39
+ MIT: 2
40
+ spdx_distinct: 10
41
+ licence_sample_check: null
42
+ unit:
43
+ pointer_level: null
44
+ pointer_count: 9503505
45
+ units: {"API response": 9240471, file: 263034}
46
+ state: null
47
+ expanded_count: null
48
+ bytes_total: 2226628535352
49
+ rows_without_size: 9501325
50
+ collections:
51
+ distinct: 10
52
+ top:
53
+ Collection: 405148
54
+ Dataset: 4137235
55
+ Image: 1598959
56
+ JournalArticle: 41897
57
+ Model: 10750
58
+ Other: 199757
59
+ PhysicalObject: 157768
60
+ Report: 58
61
+ Software: 115563
62
+ Text: 2836370
63
+ sub_collections: null
64
+ batches: null
65
+ n_ext: 38
66
+ usable_unit: null
67
+ form:
68
+ by_ext:
69
+ json: 9240474
70
+ "(none)": 260333
71
+ zip: 2228
72
+ gz: 201
73
+ pdf: 149
74
+ csv: 34
75
+ dtd: 13
76
+ txt: 11
77
+ wav: 6
78
+ p1: 2
79
+ p10: 2
80
+ p11: 2
81
+ p12: 2
82
+ p13: 2
83
+ p14: 2
84
+ p15: 2
85
+ p16: 2
86
+ p17: 2
87
+ p18: 2
88
+ p19: 2
89
+ "(18 other types)": 34
90
+ bytes_p50: 7614887
91
+ bytes_p95: 4082978219
92
+ bytes_max: 165336546177
93
+ size_quantiles:
94
+ max: 165336546177
95
+ min: 6
96
+ p25: 237652
97
+ p50: 7614887
98
+ p75: 135527039
99
+ p90: 1472308552
100
+ p95: 4082978219
101
+ p99: 16625674447
102
+ p99.9: 107422483585
103
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
104
+ by_ext_size:
105
+ - {ext: json, rows: 9240474, bytes: 50780, p50: 50780, p95: 50780, max: 50780}
106
+ - {ext: "(none)", rows: 260333, bytes: 0, p50: null, p95: null, max: null}
107
+ - {ext: zip, rows: 2228, bytes: 2226628362298, p50: 7754762, p95: 4101031245, max: 165336546177}
108
+ - {ext: gz, rows: 201, bytes: 0, p50: null, p95: null, max: null}
109
+ - {ext: pdf, rows: 149, bytes: 0, p50: null, p95: null, max: null}
110
+ - {ext: csv, rows: 34, bytes: 0, p50: null, p95: null, max: null}
111
+ - {ext: dtd, rows: 13, bytes: 122274, p50: 9289, p95: 13134, max: 13134}
112
+ - {ext: txt, rows: 11, bytes: 0, p50: null, p95: null, max: null}
113
+ - {ext: wav, rows: 6, bytes: 0, p50: null, p95: null, max: null}
114
+ - {ext: p1, rows: 2, bytes: 0, p50: null, p95: null, max: null}
115
+ - {ext: p10, rows: 2, bytes: 0, p50: null, p95: null, max: null}
116
+ - {ext: p11, rows: 2, bytes: 0, p50: null, p95: null, max: null}
117
+ - {ext: p12, rows: 2, bytes: 0, p50: null, p95: null, max: null}
118
+ - {ext: p13, rows: 2, bytes: 0, p50: null, p95: null, max: null}
119
+ - {ext: p14, rows: 2, bytes: 0, p50: null, p95: null, max: null}
120
+ - {ext: p15, rows: 2, bytes: 0, p50: null, p95: null, max: null}
121
+ - {ext: p16, rows: 2, bytes: 0, p50: null, p95: null, max: null}
122
+ - {ext: p17, rows: 2, bytes: 0, p50: null, p95: null, max: null}
123
+ - {ext: p18, rows: 2, bytes: 0, p50: null, p95: null, max: null}
124
+ - {ext: p19, rows: 2, bytes: 0, p50: null, p95: null, max: null}
125
+ over_cap:
126
+ cap: 61944279
127
+ rows_over: 698
128
+ bytes_over: 2215323225262
129
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
130
+ largest:
131
+ - pointer_id: "datacite-dois:Dataset:10.4211/hs.de74b0a457c74deca09f9a41afa03c8f/content/0/de74b0a457c74deca09f9a41afa03c8f.zip"
132
+ size_bytes: 165336546177
133
+ - pointer_id: "datacite-dois:Dataset:10.4211/hs.9462b23c5e1e46bdae6ef8abcdbed365/content/0/9462b23c5e1e46bdae6ef8abcdbed365.zip"
134
+ size_bytes: 125566506173
135
+ - pointer_id: "datacite-dois:Dataset:10.4211/hs.84d1b8b60f274ba4be155881129561a9/content/0/84d1b8b60f274ba4be155881129561a9.zip"
136
+ size_bytes: 107422483585
137
+ - pointer_id: "datacite-dois:Dataset:10.4211/hs.0f02af18e5344ae7a65dfa7fe1444f34/content/0/0f02af18e5344ae7a65dfa7fe1444f34.zip"
138
+ size_bytes: 86241451989
139
+ - pointer_id: "datacite-dois:Dataset:10.4211/hs.0d6999591fb048cab5ab71fcb690eadb/content/0/0d6999591fb048cab5ab71fcb690eadb.zip"
140
+ size_bytes: 81106041460
141
+ date_years: null
142
+ media_type: {"(none)": 263034, application/json: 9240471}
143
+ facets:
144
+ content_url_raw:
145
+ "[\\\"https://www.baidu.com/\\\"]": 1
146
+ "http://www.ngac.org.cn/Map/List": 2
147
+ public/artifacts/dblp/rdf//dblp-.nt.gz: 1
148
+ public/artifacts/dblp/xml//dblp-.dtd: 1
149
+ public/artifacts/dblp/xml//dblp-.xml.gz: 1
150
+ resource_type_general:
151
+ Collection: 405148
152
+ Dataset: 4137235
153
+ Image: 1598959
154
+ JournalArticle: 41897
155
+ Model: 10750
156
+ Other: 199757
157
+ PhysicalObject: 157768
158
+ Report: 58
159
+ Software: 115563
160
+ Text: 2836370
161
+ row_kind: {content: 263034, metadata: 9240471}
162
+ artifact: null
163
+ source_origin: null
164
+ parse_cost: null
165
+ parse_quirks: null
166
+ structure:
167
+ self_describing: null
168
+ integrity_builtin: null
169
+ has_reference_output: null
170
+ invariants: null
171
+ tools: null
172
+ ships:
173
+ problem_statement: {present: null}
174
+ reference_solution: {share: null}
175
+ tests: {present: null}
176
+ env_definition: {present: null}
177
+ known_answer: {present: null}
178
+ generator: {present: null}
179
+ upstream_labels: {present: null}
180
+ class: null
181
+ subject:
182
+ domain_span: null
183
+ shape_span: null
184
+ skill_family_span: null
185
+ ont_domain: null
186
+ tech_vocabulary: null
187
+ risk:
188
+ contamination: none
189
+ overlap_check:
190
+ done: true
191
+ pointers_flagged: 0
192
+ pointers_removed: 0
193
+ by_class: {}
194
+ benches: {}
195
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
196
+ pii: null
197
+ pii_evidence: null
198
+ malware: null
199
+ ship_bytes: unknown
200
+ subset:
201
+ rule: "take all rows; licence, size and version are labels, not filters"
202
+ rows_in_subset: 9503505
203
+ last_built: null
204
+ access:
205
+ method: null
206
+ verify: null
207
+ hosts:
208
+ "http://www.hydroshare.org": 2213
209
+ "https://api.datacite.org": 9240471
210
+ "https://columbia.redivis.com": 45165
211
+ "https://data.kigam.re.kr": 208
212
+ "https://drops.dagstuhl.de": 144
213
+ "https://huggingface.co": 10719
214
+ "https://repositorium.uni-muenster.de": 132
215
+ "https://sdss.redivis.com": 335
216
+ "https://stanford.redivis.com": 202156
217
+ "https://stanfordphs.redivis.com": 1647
218
+ uri_example: "https://api.datacite.org/dois/10.11578/1480640"
219
+ needs_auth: {"false": 9503505}
220
+ credential: null
221
+ hash_source: {computed_on_fetch: 9503505}
222
+ range_supported: null
223
+ reachable:
224
+ index_flag: {unknown: 9503505}
225
+ index_check: null
226
+ our_sample: null
227
+ drift: null
228
+ sample: null
229
+ evidence:
230
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ class: {fill: unfilled, basis: "ships not judged"}
234
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ form.by_ext:
236
+ fill: auto
237
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
238
+ checked_at: "2026-10-03"
239
+ form.date_years:
240
+ fill: auto
241
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
242
+ checked_at: "2026-10-03"
243
+ form.over_cap:
244
+ fill: auto
245
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
246
+ checked_at: "2026-10-03"
247
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
249
+ form.size_quantiles:
250
+ fill: auto
251
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
252
+ checked_at: "2026-10-03"
253
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
255
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ provenance.licence_tiers:
257
+ fill: auto
258
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
259
+ checked_at: "2026-10-03"
260
+ provenance.spdx_top:
261
+ fill: auto
262
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
263
+ checked_at: "2026-10-03"
264
+ risk.contamination:
265
+ fill: auto
266
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
267
+ checked_at: "2026-10-03"
268
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
270
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
272
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
273
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
275
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ snapshot.rev:
285
+ fill: auto
286
+ basis: "last commit listed in the publish receipt of this source"
287
+ checked_at: "2026-10-03"
288
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
290
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
293
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
294
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
295
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
296
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
297
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
298
+ subset:
299
+ fill: auto
300
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
301
+ checked_at: "2026-10-03"
302
+ unit.pointer_count:
303
+ fill: auto
304
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
305
+ checked_at: "2026-10-03"
306
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
307
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
308
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/debian-bugs.yaml ADDED
@@ -0,0 +1,1123 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: debian-bugs
8
+ parent: null
9
+ title: "Debian bug reports"
10
+ description: "Raw data of the Debian bug tracking system (bugs.debian.org) from the official rsync mirror bugs-mirror.debian.org (bts-spool-db, bts-spool-archive: open and recently closed bugs, and archived bugs). Each bug has two pointers: the full mail log of the bug (.log) and its metadata (.summary, with package, title, severity and tags)."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: a3f122af6080e58b63f8102964ed0a9830edeb44, at: "2026-10-01", rev_candidates: 2}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "No content license stated"
17
+ licence_url: "https://www.debian.org/Bugs/Access"
18
+ licence_statement: "Debian states that all bug tracker correspondence, including names and email addresses, is publicly archived (https://www.debian.org/legal/privacy). Messages remain the copyright of their senders; no content license is stated."
19
+ per_collection: null
20
+ training_ok: unknown
21
+ redistribution_ok: unknown
22
+ attribution: null
23
+ per_file: null
24
+ licence_tiers: {unknown/unknown: 2244932}
25
+ spdx_top: {"(none)": 2244932}
26
+ spdx_distinct: 1
27
+ licence_sample_check:
28
+ sample_n: 30
29
+ declared: 0
30
+ unknown_or_missing: 30
31
+ note: "mail threads carry no licence lines; the collection-level terms apply"
32
+ unit:
33
+ pointer_level: file
34
+ pointer_count: 2244932
35
+ units: {file: 2244932}
36
+ state: expanded
37
+ expanded_count: null
38
+ bytes_total: 145003200147
39
+ rows_without_size: 0
40
+ collections:
41
+ distinct: 2
42
+ top: {bts-spool-archive: 2064696, bts-spool-db: 180236}
43
+ sub_collections: null
44
+ batches: null
45
+ n_ext: 2
46
+ usable_unit: "one bug: its .log thread together with its .summary record (both share the bug number)"
47
+ form:
48
+ by_ext: {log: 1122489, summary: 1122443}
49
+ bytes_p50: 2084
50
+ bytes_p95: 150084
51
+ bytes_max: 237933861
52
+ size_quantiles:
53
+ max: 237933861
54
+ min: 147
55
+ p25: 338
56
+ p50: 2084
57
+ p75: 46670
58
+ p90: 92762
59
+ p95: 150084
60
+ p99: 576999
61
+ p99.9: 6156067
62
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
63
+ by_ext_size:
64
+ - {ext: log, rows: 1122489, bytes: 144614051838, p50: 46669, p95: 252494, max: 237933861}
65
+ - {ext: summary, rows: 1122443, bytes: 389148309, p50: 338, p95: 448, max: 18617}
66
+ over_cap:
67
+ cap: 61944279
68
+ rows_over: 53
69
+ bytes_over: 5220371103
70
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
71
+ largest:
72
+ - {pointer_id: "debian-bugs:bts-spool-archive:97/1103397.log", size_bytes: 237933861}
73
+ - {pointer_id: "debian-bugs:bts-spool-archive:76/599476.log", size_bytes: 213742133}
74
+ - {pointer_id: "debian-bugs:bts-spool-archive:18/998618.log", size_bytes: 185764695}
75
+ - {pointer_id: "debian-bugs:bts-spool-archive:15/1008115.log", size_bytes: 150002335}
76
+ - {pointer_id: "debian-bugs:bts-spool-db:69/1007169.log", size_bytes: 149957022}
77
+ date_years:
78
+ key: mtime
79
+ rows_with_date: 2244932
80
+ counts:
81
+ "1996": 4
82
+ "1997": 493
83
+ "1998": 3534
84
+ "1999": 9284
85
+ "2000": 18554
86
+ "2001": 46429
87
+ "2002": 39503
88
+ "2003": 46628
89
+ "2004": 50776
90
+ "2005": 285255
91
+ "2006": 58240
92
+ "2007": 170712
93
+ "2008": 109750
94
+ "2009": 96121
95
+ "2010": 93977
96
+ "2011": 89060
97
+ "2012": 85143
98
+ "2013": 75609
99
+ "2014": 85106
100
+ "2015": 67377
101
+ "2016": 89500
102
+ "2017": 75874
103
+ "2018": 71328
104
+ "2019": 69119
105
+ "2020": 69039
106
+ "2021": 56963
107
+ "2022": 55941
108
+ "2023": 64994
109
+ "2024": 75244
110
+ "2025": 105852
111
+ "2026": 79523
112
+ media_type: {text/plain: 2244932}
113
+ facets:
114
+ archived: {"false": 180236, "true": 2064696}
115
+ kind: {log: 1122489, summary: 1122443}
116
+ artifact: [documents-text, structured-data]
117
+ source_origin: [downloaded-source-archive]
118
+ parse_cost: needs_tool
119
+ parse_quirks:
120
+ - "mail threads are split by control-character lines and carry HTML-escaped status notes"
121
+ - "some messages say they were lost when the bug was restored from a backup"
122
+ - "a single .log can exceed 1 MB (build logs pasted in)"
123
+ structure:
124
+ self_describing: true
125
+ integrity_builtin: []
126
+ has_reference_output: false
127
+ invariants: [summary_key_value_parses, summary_format_version_present, log_and_summary_share_bug_number]
128
+ tools: [rsync, "python email"]
129
+ ships:
130
+ problem_statement: {present: true}
131
+ reference_solution:
132
+ share: {k: 6, "n": 30}
133
+ tests: {present: false, coverage: null, runs: null}
134
+ env_definition: {present: false, kind: null, runs: null}
135
+ known_answer: {present: false}
136
+ generator: {present: false}
137
+ upstream_labels:
138
+ present: true
139
+ fields: [Severity, Tags, Found-In, Fixed-In, Done, Merged-With, Package]
140
+ class: raw
141
+ subject:
142
+ domain_span: [D2.1, D2.3, D2.2]
143
+ shape_span: [S2_env_build_config, S4_data_to_artifact]
144
+ skill_family_span: [skill.systems, skill.software, skill.data]
145
+ ont_domain: [software-systems]
146
+ tech_vocabulary: [dpkg, FTBFS, FTCBFS, ITP, RFP, RM, NMU, sid, "testing migration", pkg-config, GCC-15]
147
+ risk:
148
+ contamination: none
149
+ overlap_check:
150
+ done: true
151
+ pointers_flagged: 0
152
+ pointers_removed: 0
153
+ by_class: {}
154
+ benches: {}
155
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
156
+ pii: present
157
+ pii_evidence:
158
+ sample_n: 30
159
+ files_with_email: 30
160
+ note: "From, Submitter, Done and Received headers; the archive is public"
161
+ malware: null
162
+ ship_bytes: unknown
163
+ subset:
164
+ rule: "take all rows; licence, size and version are labels, not filters"
165
+ rows_in_subset: 2244932
166
+ last_built:
167
+ date: "2026-10-02"
168
+ rule: "older rule that filtered by licence tier"
169
+ rows: 2244932
170
+ rows_in_subset: 0
171
+ excluded_by_licence: 2244932
172
+ duplicates: 0
173
+ access:
174
+ method: "rsync uri <file> (the mirror is served over rsync only)."
175
+ verify: "Compare size_bytes. The file list gives no hashes; sha256 and md5 are computed at retrieval (hash_source computed_on_fetch)."
176
+ hosts: {"rsync://bugs-mirror.debian.org": 2244932}
177
+ uri_example: "rsync://bugs-mirror.debian.org/bts-spool-archive/00/100000.log"
178
+ needs_auth: {"false": 2244932}
179
+ credential: none
180
+ hash_source: {computed_on_fetch: 2244932}
181
+ range_supported: null
182
+ reachable:
183
+ index_flag: {unknown: 2244932}
184
+ index_check:
185
+ checked_at: "2026-10-01"
186
+ sample: 30
187
+ checked: 60
188
+ ok: 58
189
+ by_status: {ok: 58, unreachable: 2}
190
+ our_sample: {checked_at: "2026-10-04", n_tried: 30, n_ok: 30, hash_checked: 30}
191
+ drift: "closed bugs are archived and rarely change; an archived bug can be unarchived and get new messages"
192
+ sample:
193
+ method: "strata are file type times the size quartile of this source (q1 to q4 by its own p25, p50, p75; unk for rows without a size); half of the samples go one per stratum to the largest strata, the rest in proportion to stratum size (largest remainder); within a stratum the rows with the smallest sha256(seed:pointer_id) are taken, so the draw is reproducible from the index"
194
+ seed: 20261004
195
+ "n": 30
196
+ n_drawn: 30
197
+ n_read: 30
198
+ unread: []
199
+ rule_n: "10 when one file type holds at least 95 percent of rows, else 30"
200
+ size_quartiles: [338, 2084, 46670]
201
+ size_filter: "only files of at most 15000000 bytes are candidates"
202
+ strata_candidates:
203
+ "summary|q1": 566914
204
+ "log|q3": 560551
205
+ "log|q4": 560421
206
+ "summary|q2": 554844
207
+ "log|q2": 709
208
+ "summary|q3": 685
209
+ strata_total: 6
210
+ strata_taken:
211
+ "log|q2": 709
212
+ "log|q3": 560551
213
+ "log|q4": 560421
214
+ "summary|q1": 566914
215
+ "summary|q2": 554844
216
+ "summary|q3": 685
217
+ fetched_at: "2026-10-04"
218
+ replaced: 0
219
+ kept: false
220
+ kept_note: "sample bytes were deleted after reading; refetch by pointer_id and compare sha256"
221
+ files:
222
+ - "no": "#00"
223
+ pointer_id: "debian-bugs:bts-spool-archive:40/10940.log"
224
+ stratum: "log|q2"
225
+ size_bytes: 1899
226
+ sha256: "8d7987872379b5fe24b67288457daaa9c21c9012e4d2585bea2616caa1316f4a"
227
+ - "no": "#01"
228
+ pointer_id: "debian-bugs:bts-spool-archive:02/163902.log"
229
+ stratum: "log|q3"
230
+ size_bytes: 19147
231
+ sha256: "9dba42da28f56c6bbb9b0745f04fa3dd7b1f56d0957e49e66e6744489b2ec2bc"
232
+ - "no": "#02"
233
+ pointer_id: "debian-bugs:bts-spool-archive:26/558726.log"
234
+ stratum: "log|q3"
235
+ size_bytes: 37615
236
+ sha256: a472362f67b2128f71d0b851361151a0f30d72963bce54234dcea10b6b9e83ff
237
+ - "no": "#03"
238
+ pointer_id: "debian-bugs:bts-spool-archive:63/901163.log"
239
+ stratum: "log|q3"
240
+ size_bytes: 45063
241
+ sha256: "8debb14681279332b9d530b0a9876f222436e15153bfd3915cd490adbd41dc32"
242
+ - "no": "#04"
243
+ pointer_id: "debian-bugs:bts-spool-archive:29/365929.log"
244
+ stratum: "log|q3"
245
+ size_bytes: 26093
246
+ sha256: "646e7bf1765432ec046d861ee57e2345ca84b86f4801bb6353f8f15872c883c4"
247
+ - "no": "#05"
248
+ pointer_id: "debian-bugs:bts-spool-archive:85/294385.log"
249
+ stratum: "log|q3"
250
+ size_bytes: 17223
251
+ sha256: da39d3aebe6921509865d4988a38f69963637c417afbc39153f41754f559c374
252
+ - "no": "#06"
253
+ pointer_id: "debian-bugs:bts-spool-archive:22/730222.log"
254
+ stratum: "log|q3"
255
+ size_bytes: 35624
256
+ sha256: "7ad34138f68e111c627def194198829ff448ebc7bf13f8b288e8a406082a38d3"
257
+ - "no": "#07"
258
+ pointer_id: "debian-bugs:bts-spool-archive:57/134657.log"
259
+ stratum: "log|q3"
260
+ size_bytes: 16586
261
+ sha256: "9b06cbf7623886113e9fddaaadfc71dc64a52ed804371d5249fa560df3e56acd"
262
+ - "no": "#08"
263
+ pointer_id: "debian-bugs:bts-spool-archive:46/352046.log"
264
+ stratum: "log|q4"
265
+ size_bytes: 48465
266
+ sha256: "455d19a23e476afac8b7bea5be1378e22b9871787a8b56d8697fc3b21799fa59"
267
+ - "no": "#09"
268
+ pointer_id: "debian-bugs:bts-spool-archive:54/970454.log"
269
+ stratum: "log|q4"
270
+ size_bytes: 1058118
271
+ sha256: "9c842ef1a479498566458dd094faca5205ac589106972d7a695cf5273a26312a"
272
+ - "no": "#10"
273
+ pointer_id: "debian-bugs:bts-spool-archive:94/959494.log"
274
+ stratum: "log|q4"
275
+ size_bytes: 116144
276
+ sha256: bcd3ccdefd92a1570bdfbd6da0000f0840fc0338e63a5ac8418f94901801a2a5
277
+ - "no": "#11"
278
+ pointer_id: "debian-bugs:bts-spool-archive:49/1084249.log"
279
+ stratum: "log|q4"
280
+ size_bytes: 134416
281
+ sha256: abd9daf61be1b22afe1a6d627c331911e3e8745577c38ec12e702d6d072b9251
282
+ - "no": "#12"
283
+ pointer_id: "debian-bugs:bts-spool-archive:24/1096524.log"
284
+ stratum: "log|q4"
285
+ size_bytes: 578692
286
+ sha256: ff237979409dc55b42faed94e4776479c14de906cc63bba90e5f1ba7f9032d57
287
+ - "no": "#13"
288
+ pointer_id: "debian-bugs:bts-spool-archive:51/562851.log"
289
+ stratum: "log|q4"
290
+ size_bytes: 55416
291
+ sha256: "6f2cb3c134ccf23396356fede165833c996e89744dbe06190c8b760cc5add335"
292
+ - "no": "#14"
293
+ pointer_id: "debian-bugs:bts-spool-archive:81/238481.log"
294
+ stratum: "log|q4"
295
+ size_bytes: 46732
296
+ sha256: "9c4d49a55258503c89fa0d59f917c63f77e0c36d4cff96daaec88f19c1fe3771"
297
+ - "no": "#15"
298
+ pointer_id: "debian-bugs:bts-spool-archive:86/908086.summary"
299
+ stratum: "summary|q1"
300
+ size_bytes: 331
301
+ sha256: "49938086036cdec6331dffc794ce9b7a0cb4037d8ff89d06208337519f375c6d"
302
+ - "no": "#16"
303
+ pointer_id: "debian-bugs:bts-spool-archive:00/90700.summary"
304
+ stratum: "summary|q1"
305
+ size_bytes: 328
306
+ sha256: a5eada651a032ab541b1b50834ff8aa3f4506d6375bb56fad98cb236b09aa487
307
+ - "no": "#17"
308
+ pointer_id: "debian-bugs:bts-spool-archive:67/106867.summary"
309
+ stratum: "summary|q1"
310
+ size_bytes: 281
311
+ sha256: a117f487004c510ea7dfea3297f5fa6a29464563e592974ab32b283c09d5b718
312
+ - "no": "#18"
313
+ pointer_id: "debian-bugs:bts-spool-archive:49/803549.summary"
314
+ stratum: "summary|q1"
315
+ size_bytes: 331
316
+ sha256: e06aa4ba7e8aca26ddb26fb9423db2cc890e8c9c2d9556508590a4c4f4ac6436
317
+ - "no": "#19"
318
+ pointer_id: "debian-bugs:bts-spool-archive:11/230011.summary"
319
+ stratum: "summary|q1"
320
+ size_bytes: 323
321
+ sha256: "0181d797d74e1a93060cd115b2221e45195b1cabc8b2ce171c0cafbe8a193837"
322
+ - "no": "#20"
323
+ pointer_id: "debian-bugs:bts-spool-archive:15/11115.summary"
324
+ stratum: "summary|q1"
325
+ size_bytes: 251
326
+ sha256: b36fe607cd69490a4eabfc04f638fc49cecb178a8344f9d1580978863ac81e59
327
+ - "no": "#21"
328
+ pointer_id: "debian-bugs:bts-spool-archive:84/694584.summary"
329
+ stratum: "summary|q1"
330
+ size_bytes: 317
331
+ sha256: "1b0435c30a6048fdb3ab5602031a3525eed5ebd9b709693a0f37ce70e00301dc"
332
+ - "no": "#22"
333
+ pointer_id: "debian-bugs:bts-spool-archive:92/956492.summary"
334
+ stratum: "summary|q2"
335
+ size_bytes: 348
336
+ sha256: "50faada02750a14dbad211e3d4d2f1d69bff61e9ae0fb406b95644f6c00b33b1"
337
+ - "no": "#23"
338
+ pointer_id: "debian-bugs:bts-spool-archive:08/384308.summary"
339
+ stratum: "summary|q2"
340
+ size_bytes: 430
341
+ sha256: a1afec217a932594e784c218a15ef4c606fcd83d9914962753c80f4ad0804a85
342
+ - "no": "#24"
343
+ pointer_id: "debian-bugs:bts-spool-archive:06/290706.summary"
344
+ stratum: "summary|q2"
345
+ size_bytes: 366
346
+ sha256: "591a639c2561651161c273f99b1930b4ca767825ba3f7e99b01305de32103469"
347
+ - "no": "#25"
348
+ pointer_id: "debian-bugs:bts-spool-archive:21/557521.summary"
349
+ stratum: "summary|q2"
350
+ size_bytes: 357
351
+ sha256: b709e8def929c15bd6726bad907fb2011512a5c26aef7652d92f83fcae554017
352
+ - "no": "#26"
353
+ pointer_id: "debian-bugs:bts-spool-archive:02/242902.summary"
354
+ stratum: "summary|q2"
355
+ size_bytes: 402
356
+ sha256: "093988340772e4a0f0aa406223f1c848d2d1d46a03bce182bf0c61eeefa6466d"
357
+ - "no": "#27"
358
+ pointer_id: "debian-bugs:bts-spool-archive:52/1081452.summary"
359
+ stratum: "summary|q2"
360
+ size_bytes: 385
361
+ sha256: "2a502d015c13292487fc45af3d1bb980a51a874a342452e624372eab2eca7065"
362
+ - "no": "#28"
363
+ pointer_id: "debian-bugs:bts-spool-archive:22/595122.summary"
364
+ stratum: "summary|q2"
365
+ size_bytes: 348
366
+ sha256: "48757a41f7cbf7f5139b2b1e74d966a827fd93db7d2f6b24941187c95278e1ac"
367
+ - "no": "#29"
368
+ pointer_id: "debian-bugs:bts-spool-archive:69/371569.summary"
369
+ stratum: "summary|q3"
370
+ size_bytes: 2677
371
+ sha256: "526bde569c0b280b12adf3eb0a731365d9591f49e9ef782283ed8d5dfdd16557"
372
+ evidence:
373
+ access.credential:
374
+ fill: judgment
375
+ basis: "all 30 fetched anonymously over HTTPS"
376
+ files:
377
+ - "#00"
378
+ - "#01"
379
+ - "#02"
380
+ - "#03"
381
+ - "#04"
382
+ - "#05"
383
+ - "#06"
384
+ - "#07"
385
+ - "#08"
386
+ - "#09"
387
+ - "#10"
388
+ - "#11"
389
+ - "#12"
390
+ - "#13"
391
+ - "#14"
392
+ - "#15"
393
+ - "#16"
394
+ - "#17"
395
+ - "#18"
396
+ - "#19"
397
+ - "#20"
398
+ - "#21"
399
+ - "#22"
400
+ - "#23"
401
+ - "#24"
402
+ - "#25"
403
+ - "#26"
404
+ - "#27"
405
+ - "#28"
406
+ - "#29"
407
+ sample_n: 30
408
+ checked_at: "2026-10-04"
409
+ access.drift:
410
+ fill: judgment
411
+ basis: "the archive spool covers 1997 to 2025 in this sample (#00 1997, #12 2025)"
412
+ files: ["#00", "#12"]
413
+ sample_n: 30
414
+ checked_at: "2026-10-04"
415
+ access.range_supported: {fill: unfilled, basis: "no sample evidence given for this field"}
416
+ class:
417
+ fill: derived
418
+ basis: "rule over ships.tests, env_definition, known_answer, generator and structure; manifest_not_run when tests and a dependency manifest are present but the build was not run"
419
+ form.artifact:
420
+ fill: judgment
421
+ basis: ".log is mail text with HTML-escaped headers; .summary is key: value lines; not logs-traces because these are not system event logs"
422
+ files: ["#00", "#22"]
423
+ sample_n: 30
424
+ checked_at: "2026-10-04"
425
+ form.by_ext:
426
+ fill: auto
427
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
428
+ checked_at: "2026-10-03"
429
+ form.date_years:
430
+ fill: auto
431
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
432
+ checked_at: "2026-10-03"
433
+ form.over_cap:
434
+ fill: auto
435
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
436
+ checked_at: "2026-10-03"
437
+ form.parse_cost:
438
+ fill: judgment
439
+ basis: "file(1) says data for all 15 .log samples: control-character separators and HTML-escaped mail headers need a parser; .summary is plain text"
440
+ files:
441
+ - "#00"
442
+ - "#01"
443
+ - "#02"
444
+ - "#03"
445
+ - "#04"
446
+ - "#05"
447
+ - "#06"
448
+ - "#07"
449
+ - "#08"
450
+ - "#09"
451
+ - "#10"
452
+ - "#11"
453
+ - "#12"
454
+ - "#13"
455
+ - "#14"
456
+ sample_n: 30
457
+ checked_at: "2026-10-04"
458
+ form.parse_quirks:
459
+ fill: judgment
460
+ basis: "#00 shows the backup-loss notes; #09 is 1,058,118 bytes"
461
+ files: ["#00", "#09", "#12"]
462
+ sample_n: 30
463
+ checked_at: "2026-10-04"
464
+ form.size_quantiles:
465
+ fill: auto
466
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
467
+ checked_at: "2026-10-03"
468
+ form.source_origin:
469
+ fill: judgment
470
+ basis: "each report is filed against a Debian source package (Package: and Source: lines, version in Version: or Found:); the code a task would build on is fetched as that source package (dsc and orig tarball)"
471
+ files:
472
+ - "#00"
473
+ - "#01"
474
+ - "#02"
475
+ - "#03"
476
+ - "#04"
477
+ - "#05"
478
+ - "#06"
479
+ - "#07"
480
+ - "#08"
481
+ - "#09"
482
+ - "#10"
483
+ - "#11"
484
+ - "#12"
485
+ - "#13"
486
+ - "#14"
487
+ - "#15"
488
+ - "#16"
489
+ - "#17"
490
+ - "#18"
491
+ - "#19"
492
+ - "#20"
493
+ - "#21"
494
+ - "#22"
495
+ - "#23"
496
+ - "#24"
497
+ - "#25"
498
+ - "#26"
499
+ - "#27"
500
+ - "#28"
501
+ - "#29"
502
+ sample_n: 30
503
+ checked_at: "2026-10-04"
504
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
505
+ provenance.licence_sample_check:
506
+ fill: judgment
507
+ basis: "no licence text in any sample"
508
+ files:
509
+ - "#00"
510
+ - "#01"
511
+ - "#02"
512
+ - "#03"
513
+ - "#04"
514
+ - "#05"
515
+ - "#06"
516
+ - "#07"
517
+ - "#08"
518
+ - "#09"
519
+ - "#10"
520
+ - "#11"
521
+ - "#12"
522
+ - "#13"
523
+ - "#14"
524
+ - "#15"
525
+ - "#16"
526
+ - "#17"
527
+ - "#18"
528
+ - "#19"
529
+ - "#20"
530
+ - "#21"
531
+ - "#22"
532
+ - "#23"
533
+ - "#24"
534
+ - "#25"
535
+ - "#26"
536
+ - "#27"
537
+ - "#28"
538
+ - "#29"
539
+ sample_n: 30
540
+ checked_at: "2026-10-04"
541
+ provenance.licence_tiers:
542
+ fill: auto
543
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
544
+ checked_at: "2026-10-03"
545
+ provenance.spdx_top:
546
+ fill: auto
547
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
548
+ checked_at: "2026-10-03"
549
+ risk.contamination:
550
+ fill: auto
551
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
552
+ checked_at: "2026-10-03"
553
+ risk.malware: {fill: unfilled, basis: "no sample evidence given for this field"}
554
+ risk.pii:
555
+ fill: judgment
556
+ basis: "names and email addresses of reporters and maintainers in all 30 samples (every digest has 2 to 13 addresses); mail Received headers show hosts and IP addresses (#00); re-checked under the WP-T-44c definition (a user name on a platform counts as present)"
557
+ files:
558
+ - "#00"
559
+ - "#01"
560
+ - "#02"
561
+ - "#03"
562
+ - "#04"
563
+ - "#05"
564
+ - "#06"
565
+ - "#07"
566
+ - "#08"
567
+ - "#09"
568
+ - "#10"
569
+ - "#11"
570
+ - "#12"
571
+ - "#13"
572
+ - "#14"
573
+ - "#15"
574
+ - "#16"
575
+ - "#17"
576
+ - "#18"
577
+ - "#19"
578
+ - "#20"
579
+ - "#21"
580
+ - "#22"
581
+ - "#23"
582
+ - "#24"
583
+ - "#25"
584
+ - "#26"
585
+ - "#27"
586
+ - "#28"
587
+ - "#29"
588
+ sample_n: 30
589
+ checked_at: "2026-10-04"
590
+ risk.pii_evidence:
591
+ fill: judgment
592
+ basis: "email regex over the digests"
593
+ files:
594
+ - "#00"
595
+ - "#01"
596
+ - "#02"
597
+ - "#03"
598
+ - "#04"
599
+ - "#05"
600
+ - "#06"
601
+ - "#07"
602
+ - "#08"
603
+ - "#09"
604
+ - "#10"
605
+ - "#11"
606
+ - "#12"
607
+ - "#13"
608
+ - "#14"
609
+ - "#15"
610
+ - "#16"
611
+ - "#17"
612
+ - "#18"
613
+ - "#19"
614
+ - "#20"
615
+ - "#21"
616
+ - "#22"
617
+ - "#23"
618
+ - "#24"
619
+ - "#25"
620
+ - "#26"
621
+ - "#27"
622
+ - "#28"
623
+ - "#29"
624
+ sample_n: 30
625
+ checked_at: "2026-10-04"
626
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
627
+ ships.env_definition.kind: {fill: unfilled, basis: "no sample evidence given for this field"}
628
+ ships.env_definition.present:
629
+ fill: judgment
630
+ basis: "no Dockerfile or manifest; Found-In names a package version only"
631
+ files:
632
+ - "#00"
633
+ - "#01"
634
+ - "#02"
635
+ - "#03"
636
+ - "#04"
637
+ - "#05"
638
+ - "#06"
639
+ - "#07"
640
+ - "#08"
641
+ - "#09"
642
+ - "#10"
643
+ - "#11"
644
+ - "#12"
645
+ - "#13"
646
+ - "#14"
647
+ - "#15"
648
+ - "#16"
649
+ - "#17"
650
+ - "#18"
651
+ - "#19"
652
+ - "#20"
653
+ - "#21"
654
+ - "#22"
655
+ - "#23"
656
+ - "#24"
657
+ - "#25"
658
+ - "#26"
659
+ - "#27"
660
+ - "#28"
661
+ - "#29"
662
+ sample_n: 30
663
+ checked_at: "2026-10-04"
664
+ ships.env_definition.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
665
+ ships.generator.present:
666
+ fill: judgment
667
+ basis: "nothing produces (input, answer) pairs"
668
+ files:
669
+ - "#00"
670
+ - "#01"
671
+ - "#02"
672
+ - "#03"
673
+ - "#04"
674
+ - "#05"
675
+ - "#06"
676
+ - "#07"
677
+ - "#08"
678
+ - "#09"
679
+ - "#10"
680
+ - "#11"
681
+ - "#12"
682
+ - "#13"
683
+ - "#14"
684
+ - "#15"
685
+ - "#16"
686
+ - "#17"
687
+ - "#18"
688
+ - "#19"
689
+ - "#20"
690
+ - "#21"
691
+ - "#22"
692
+ - "#23"
693
+ - "#24"
694
+ - "#25"
695
+ - "#26"
696
+ - "#27"
697
+ - "#28"
698
+ - "#29"
699
+ sample_n: 30
700
+ checked_at: "2026-10-04"
701
+ ships.known_answer.present:
702
+ fill: judgment
703
+ basis: "Fixed-In gives the fixing version but no checkable answer value"
704
+ files:
705
+ - "#15"
706
+ - "#16"
707
+ - "#17"
708
+ - "#18"
709
+ - "#19"
710
+ - "#20"
711
+ - "#21"
712
+ - "#22"
713
+ - "#23"
714
+ - "#24"
715
+ - "#25"
716
+ - "#26"
717
+ - "#27"
718
+ - "#28"
719
+ - "#29"
720
+ sample_n: 30
721
+ checked_at: "2026-10-04"
722
+ ships.problem_statement.present:
723
+ fill: judgment
724
+ basis: "the first message of every .log is a bug report with Subject, Package and Version (#00 passwd, #03 FTCBFS, #10 cmake)"
725
+ files: ["#00", "#03", "#10"]
726
+ sample_n: 30
727
+ checked_at: "2026-10-04"
728
+ ships.reference_solution.share:
729
+ fill: judgment
730
+ basis: "6 of 30 sample files read: diff hunks or patch attachments in #03, #04, #06 and Tags: patch in #22, #24, #26"
731
+ files: ["#03", "#04", "#06", "#22", "#24", "#26"]
732
+ sample_n: 30
733
+ checked_at: "2026-10-04"
734
+ ships.tests.coverage: {fill: unfilled, basis: "no sample evidence given for this field"}
735
+ ships.tests.present:
736
+ fill: judgment
737
+ basis: "no test files; FTBFS reports quote failing test output but do not ship tests"
738
+ files:
739
+ - "#00"
740
+ - "#01"
741
+ - "#02"
742
+ - "#03"
743
+ - "#04"
744
+ - "#05"
745
+ - "#06"
746
+ - "#07"
747
+ - "#08"
748
+ - "#09"
749
+ - "#10"
750
+ - "#11"
751
+ - "#12"
752
+ - "#13"
753
+ - "#14"
754
+ - "#15"
755
+ - "#16"
756
+ - "#17"
757
+ - "#18"
758
+ - "#19"
759
+ - "#20"
760
+ - "#21"
761
+ - "#22"
762
+ - "#23"
763
+ - "#24"
764
+ - "#25"
765
+ - "#26"
766
+ - "#27"
767
+ - "#28"
768
+ - "#29"
769
+ sample_n: 30
770
+ checked_at: "2026-10-04"
771
+ ships.tests.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
772
+ ships.upstream_labels.fields:
773
+ fill: judgment
774
+ basis: "keys seen in the summaries (#22, #26, #29)"
775
+ files: ["#22", "#26", "#29"]
776
+ sample_n: 30
777
+ checked_at: "2026-10-04"
778
+ ships.upstream_labels.present:
779
+ fill: judgment
780
+ basis: "Severity in all 15 summaries, Done in all 15, Fixed-In in 10, Tags in 6"
781
+ files:
782
+ - "#15"
783
+ - "#16"
784
+ - "#17"
785
+ - "#18"
786
+ - "#19"
787
+ - "#20"
788
+ - "#21"
789
+ - "#22"
790
+ - "#23"
791
+ - "#24"
792
+ - "#25"
793
+ - "#26"
794
+ - "#27"
795
+ - "#28"
796
+ - "#29"
797
+ sample_n: 30
798
+ checked_at: "2026-10-04"
799
+ snapshot.rev:
800
+ fill: auto
801
+ basis: "last commit listed in the publish receipt of this source"
802
+ checked_at: "2026-10-03"
803
+ structure.has_reference_output:
804
+ fill: judgment
805
+ basis: "no expected output file"
806
+ files:
807
+ - "#00"
808
+ - "#01"
809
+ - "#02"
810
+ - "#03"
811
+ - "#04"
812
+ - "#05"
813
+ - "#06"
814
+ - "#07"
815
+ - "#08"
816
+ - "#09"
817
+ - "#10"
818
+ - "#11"
819
+ - "#12"
820
+ - "#13"
821
+ - "#14"
822
+ - "#15"
823
+ - "#16"
824
+ - "#17"
825
+ - "#18"
826
+ - "#19"
827
+ - "#20"
828
+ - "#21"
829
+ - "#22"
830
+ - "#23"
831
+ - "#24"
832
+ - "#25"
833
+ - "#26"
834
+ - "#27"
835
+ - "#28"
836
+ - "#29"
837
+ sample_n: 30
838
+ checked_at: "2026-10-04"
839
+ structure.integrity_builtin:
840
+ fill: judgment
841
+ basis: "no checksum or signature inside the files"
842
+ files:
843
+ - "#00"
844
+ - "#01"
845
+ - "#02"
846
+ - "#03"
847
+ - "#04"
848
+ - "#05"
849
+ - "#06"
850
+ - "#07"
851
+ - "#08"
852
+ - "#09"
853
+ - "#10"
854
+ - "#11"
855
+ - "#12"
856
+ - "#13"
857
+ - "#14"
858
+ - "#15"
859
+ - "#16"
860
+ - "#17"
861
+ - "#18"
862
+ - "#19"
863
+ - "#20"
864
+ - "#21"
865
+ - "#22"
866
+ - "#23"
867
+ - "#24"
868
+ - "#25"
869
+ - "#26"
870
+ - "#27"
871
+ - "#28"
872
+ - "#29"
873
+ sample_n: 30
874
+ checked_at: "2026-10-04"
875
+ structure.invariants:
876
+ fill: judgment
877
+ basis: "all 15 summaries are key: value lines with Format-Version"
878
+ files:
879
+ - "#15"
880
+ - "#16"
881
+ - "#17"
882
+ - "#18"
883
+ - "#19"
884
+ - "#20"
885
+ - "#21"
886
+ - "#22"
887
+ - "#23"
888
+ - "#24"
889
+ - "#25"
890
+ - "#26"
891
+ - "#27"
892
+ - "#28"
893
+ - "#29"
894
+ sample_n: 30
895
+ checked_at: "2026-10-04"
896
+ structure.self_describing:
897
+ fill: judgment
898
+ basis: ".summary starts with Format-Version (2 or 3) and gives Package, Severity, Found-In, Fixed-In, Done as named keys"
899
+ files:
900
+ - "#15"
901
+ - "#16"
902
+ - "#17"
903
+ - "#18"
904
+ - "#19"
905
+ - "#20"
906
+ - "#21"
907
+ - "#22"
908
+ - "#23"
909
+ - "#24"
910
+ - "#25"
911
+ - "#26"
912
+ - "#27"
913
+ - "#28"
914
+ - "#29"
915
+ sample_n: 30
916
+ checked_at: "2026-10-04"
917
+ structure.tools:
918
+ fill: judgment
919
+ basis: "the spool is mirrored with rsync; the mail parts parse with an email parser"
920
+ files:
921
+ - "#00"
922
+ - "#01"
923
+ - "#02"
924
+ - "#03"
925
+ - "#04"
926
+ - "#05"
927
+ - "#06"
928
+ - "#07"
929
+ - "#08"
930
+ - "#09"
931
+ - "#10"
932
+ - "#11"
933
+ - "#12"
934
+ - "#13"
935
+ - "#14"
936
+ - "#15"
937
+ - "#16"
938
+ - "#17"
939
+ - "#18"
940
+ - "#19"
941
+ - "#20"
942
+ - "#21"
943
+ - "#22"
944
+ - "#23"
945
+ - "#24"
946
+ - "#25"
947
+ - "#26"
948
+ - "#27"
949
+ - "#28"
950
+ - "#29"
951
+ sample_n: 30
952
+ checked_at: "2026-10-04"
953
+ subject.domain_span:
954
+ fill: judgment
955
+ basis: "D2.1 build failures (#03, #09 to #12, #15, #16, #26, #29); D2.2 service and hardware configuration (#06, #14, #20, #22); D2.3 packaging requests and removals (#01, #17, #18, #25, #27); D2.6 diagnosis (#13 load average); D8.4 buffer overflow report (#24); trimmed to at most three main values (WP-T-44c): kept the values most samples show, dropped a parent code when its subcode is kept"
956
+ files: ["#01", "#03", "#06", "#09", "#13", "#14", "#17", "#24", "#29"]
957
+ sample_n: 30
958
+ checked_at: "2026-10-04"
959
+ subject.ont_domain:
960
+ fill: judgment
961
+ basis: "all 30 samples are about Debian packages or Debian infrastructure"
962
+ files:
963
+ - "#00"
964
+ - "#01"
965
+ - "#02"
966
+ - "#03"
967
+ - "#04"
968
+ - "#05"
969
+ - "#06"
970
+ - "#07"
971
+ - "#08"
972
+ - "#09"
973
+ - "#10"
974
+ - "#11"
975
+ - "#12"
976
+ - "#13"
977
+ - "#14"
978
+ - "#15"
979
+ - "#16"
980
+ - "#17"
981
+ - "#18"
982
+ - "#19"
983
+ - "#20"
984
+ - "#21"
985
+ - "#22"
986
+ - "#23"
987
+ - "#24"
988
+ - "#25"
989
+ - "#26"
990
+ - "#27"
991
+ - "#28"
992
+ - "#29"
993
+ sample_n: 30
994
+ checked_at: "2026-10-04"
995
+ subject.shape_span:
996
+ fill: judgment
997
+ basis: "build-failure reports with fixes can seed build tasks (#03, #10, #26); the threads can seed triage and summary tasks"
998
+ files: ["#03", "#10", "#26"]
999
+ sample_n: 30
1000
+ checked_at: "2026-10-04"
1001
+ subject.skill_family_span:
1002
+ fill: judgment
1003
+ basis: "Debian packaging and system administration; patches to code; reading long threads"
1004
+ files:
1005
+ - "#00"
1006
+ - "#01"
1007
+ - "#02"
1008
+ - "#03"
1009
+ - "#04"
1010
+ - "#05"
1011
+ - "#06"
1012
+ - "#07"
1013
+ - "#08"
1014
+ - "#09"
1015
+ - "#10"
1016
+ - "#11"
1017
+ - "#12"
1018
+ - "#13"
1019
+ - "#14"
1020
+ - "#15"
1021
+ - "#16"
1022
+ - "#17"
1023
+ - "#18"
1024
+ - "#19"
1025
+ - "#20"
1026
+ - "#21"
1027
+ - "#22"
1028
+ - "#23"
1029
+ - "#24"
1030
+ - "#25"
1031
+ - "#26"
1032
+ - "#27"
1033
+ - "#28"
1034
+ - "#29"
1035
+ sample_n: 30
1036
+ checked_at: "2026-10-04"
1037
+ subject.tech_vocabulary:
1038
+ fill: judgment
1039
+ basis: "terms in the subjects and bodies"
1040
+ files: ["#01", "#03", "#12", "#17", "#18", "#22", "#27"]
1041
+ sample_n: 30
1042
+ checked_at: "2026-10-04"
1043
+ subset:
1044
+ fill: auto
1045
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
1046
+ checked_at: "2026-10-03"
1047
+ unit.pointer_count:
1048
+ fill: auto
1049
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
1050
+ checked_at: "2026-10-03"
1051
+ unit.pointer_level:
1052
+ fill: judgment
1053
+ basis: "each pointer is one .log (full mail thread of one bug) or one .summary (status record of one bug)"
1054
+ files:
1055
+ - "#00"
1056
+ - "#01"
1057
+ - "#02"
1058
+ - "#03"
1059
+ - "#04"
1060
+ - "#05"
1061
+ - "#06"
1062
+ - "#07"
1063
+ - "#08"
1064
+ - "#09"
1065
+ - "#10"
1066
+ - "#11"
1067
+ - "#12"
1068
+ - "#13"
1069
+ - "#14"
1070
+ - "#15"
1071
+ - "#16"
1072
+ - "#17"
1073
+ - "#18"
1074
+ - "#19"
1075
+ - "#20"
1076
+ - "#21"
1077
+ - "#22"
1078
+ - "#23"
1079
+ - "#24"
1080
+ - "#25"
1081
+ - "#26"
1082
+ - "#27"
1083
+ - "#28"
1084
+ - "#29"
1085
+ sample_n: 30
1086
+ checked_at: "2026-10-04"
1087
+ unit.state: {fill: derived, basis: "packed when pointer_level is archive, else expanded"}
1088
+ unit.usable_unit:
1089
+ fill: judgment
1090
+ basis: "15 .log and 15 .summary samples; each names one bug"
1091
+ files:
1092
+ - "#00"
1093
+ - "#01"
1094
+ - "#02"
1095
+ - "#03"
1096
+ - "#04"
1097
+ - "#05"
1098
+ - "#06"
1099
+ - "#07"
1100
+ - "#08"
1101
+ - "#09"
1102
+ - "#10"
1103
+ - "#11"
1104
+ - "#12"
1105
+ - "#13"
1106
+ - "#14"
1107
+ - "#15"
1108
+ - "#16"
1109
+ - "#17"
1110
+ - "#18"
1111
+ - "#19"
1112
+ - "#20"
1113
+ - "#21"
1114
+ - "#22"
1115
+ - "#23"
1116
+ - "#24"
1117
+ - "#25"
1118
+ - "#26"
1119
+ - "#27"
1120
+ - "#28"
1121
+ - "#29"
1122
+ sample_n: 30
1123
+ checked_at: "2026-10-04"
cards/debian-sources.yaml ADDED
@@ -0,0 +1,235 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: debian-sources
8
+ parent: null
9
+ title: "Debian source packages"
10
+ description: "All source files (.dsc, upstream tarballs and Debian tarballs) of Debian source packages (trixie; main). Lists and hashes come from Sources.xz on the mirror network, checked against Release first."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: "75f20949683e4dacf469d9c0c516d87dfe7ebc92", at: "2026-10-01", rev_candidates: 2}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "Debian Free Software Guidelines (main)"
17
+ licence_url: "https://www.debian.org/social_contract#guidelines"
18
+ licence_statement: "Everything in main meets the Debian Free Software Guidelines; per-package licenses are in debian/copyright."
19
+ per_collection: null
20
+ training_ok: "yes"
21
+ redistribution_ok: conditional
22
+ attribution: null
23
+ per_file: "Rows from contrib or non-free carry license_override describing that area."
24
+ licence_tiers: {yes/conditional: 115614}
25
+ spdx_top: {"(none)": 115614}
26
+ spdx_distinct: 1
27
+ licence_sample_check: null
28
+ unit:
29
+ pointer_level: null
30
+ pointer_count: 115614
31
+ units: {file: 115614}
32
+ state: null
33
+ expanded_count: null
34
+ bytes_total: 99187805300
35
+ rows_without_size: 0
36
+ collections:
37
+ distinct: 37591
38
+ top:
39
+ firefox-esr: 105
40
+ node-babel7: 42
41
+ node-deep-equal: 35
42
+ node-gulp: 36
43
+ node-istanbul: 33
44
+ node-jest: 28
45
+ node-jsdom: 32
46
+ node-lodash: 292
47
+ node-stdlib: 43
48
+ sphinx: 50
49
+ sub_collections: null
50
+ batches: null
51
+ n_ext: 5
52
+ usable_unit: null
53
+ form:
54
+ by_ext: {xz: 46917, dsc: 37643, gz: 28265, asc: 1544, bz2: 1245}
55
+ bytes_p50: 3796
56
+ bytes_p95: 1095916
57
+ bytes_max: 2840925236
58
+ size_quantiles:
59
+ max: 2840925236
60
+ min: 65
61
+ p25: 2295
62
+ p50: 3796
63
+ p75: 23456
64
+ p90: 273640
65
+ p95: 1095916
66
+ p99: 11782775
67
+ p99.9: 130334635
68
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
69
+ by_ext_size:
70
+ - {ext: xz, rows: 46917, bytes: 53000735708, p50: 5128, p95: 1155176, max: 2840925236}
71
+ - {ext: dsc, rows: 37643, bytes: 87718299, p50: 2201, p95: 2992, max: 290418}
72
+ - {ext: gz, rows: 28265, bytes: 40910117865, p50: 42489, p95: 3570573, max: 1750163980}
73
+ - {ext: asc, rows: 1544, bytes: 2356438, p50: 833, p95: 862, max: 1089270}
74
+ - {ext: bz2, rows: 1245, bytes: 5186876990, p50: 324373, p95: 10937641, max: 461916727}
75
+ over_cap:
76
+ cap: 61944279
77
+ rows_over: 261
78
+ bytes_over: 54317677418
79
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
80
+ largest:
81
+ - pointer_id: "debian-sources:texlive-extra:texlive-extra_2024.20250309.orig.tar.xz"
82
+ size_bytes: 2840925236
83
+ - {pointer_id: "debian-sources:unidic-mecab:unidic-mecab_3.1.1.orig.tar.gz", size_bytes: 1750163980}
84
+ - pointer_id: "debian-sources:sagemath-database-cremona-elliptic-curves:sagemath-database-cremona-elliptic-curves_20221013.orig.tar.gz"
85
+ size_bytes: 1679624245
86
+ - {pointer_id: "debian-sources:0ad-data:0ad-data_0.27.0.orig.tar.xz", size_bytes: 1367701984}
87
+ - pointer_id: "debian-sources:flightgear-data:flightgear-data_2020.3.19+dfsg.orig.tar.xz"
88
+ size_bytes: 1346965760
89
+ date_years: null
90
+ media_type: {"(none)": 115614}
91
+ facets:
92
+ format: {"1.0": 493, "3.0 (native)": 1496, "3.0 (quilt)": 113625}
93
+ artifact: null
94
+ source_origin: null
95
+ parse_cost: null
96
+ parse_quirks: null
97
+ structure:
98
+ self_describing: null
99
+ integrity_builtin: null
100
+ has_reference_output: null
101
+ invariants: null
102
+ tools: null
103
+ ships:
104
+ problem_statement: {present: null}
105
+ reference_solution: {share: null}
106
+ tests: {present: null}
107
+ env_definition: {present: null}
108
+ known_answer: {present: null}
109
+ generator: {present: null}
110
+ upstream_labels: {present: null}
111
+ class: null
112
+ subject:
113
+ domain_span: null
114
+ shape_span: null
115
+ skill_family_span: null
116
+ ont_domain: null
117
+ tech_vocabulary: null
118
+ risk:
119
+ contamination: none
120
+ overlap_check:
121
+ done: true
122
+ pointers_flagged: 0
123
+ pointers_removed: 0
124
+ by_class: {}
125
+ benches: {}
126
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
127
+ pii: null
128
+ pii_evidence: null
129
+ malware: null
130
+ ship_bytes: conditional
131
+ subset:
132
+ rule: "take all rows; licence, size and version are labels, not filters"
133
+ rows_in_subset: 115614
134
+ last_built:
135
+ date: "2026-10-02"
136
+ rule: "older rule that filtered by licence tier"
137
+ rows: 115614
138
+ rows_in_subset: 114298
139
+ excluded_by_licence: 0
140
+ duplicates: 1316
141
+ access:
142
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
143
+ verify: "Compare size_bytes, sha256 and md5 (Checksums-Sha256 and Files in Sources)."
144
+ hosts: {"https://deb.debian.org": 115614}
145
+ uri_example: "https://deb.debian.org/debian/pool/main/0/0ad-data/0ad-data_0.27.0-1.debian.tar.xz"
146
+ needs_auth: {"false": 115614}
147
+ credential: null
148
+ hash_source: {source_published: 115614}
149
+ range_supported: null
150
+ reachable:
151
+ index_flag: {unknown: 115614}
152
+ index_check: null
153
+ our_sample: null
154
+ drift: null
155
+ sample: null
156
+ evidence:
157
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
158
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
159
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
160
+ class: {fill: unfilled, basis: "ships not judged"}
161
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
162
+ form.by_ext:
163
+ fill: auto
164
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
165
+ checked_at: "2026-10-03"
166
+ form.date_years:
167
+ fill: auto
168
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
169
+ checked_at: "2026-10-03"
170
+ form.over_cap:
171
+ fill: auto
172
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
173
+ checked_at: "2026-10-03"
174
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
175
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
176
+ form.size_quantiles:
177
+ fill: auto
178
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
179
+ checked_at: "2026-10-03"
180
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
181
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
182
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ provenance.licence_tiers:
184
+ fill: auto
185
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
186
+ checked_at: "2026-10-03"
187
+ provenance.spdx_top:
188
+ fill: auto
189
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
190
+ checked_at: "2026-10-03"
191
+ risk.contamination:
192
+ fill: auto
193
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
194
+ checked_at: "2026-10-03"
195
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
196
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
197
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
198
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
199
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ snapshot.rev:
212
+ fill: auto
213
+ basis: "last commit listed in the publish receipt of this source"
214
+ checked_at: "2026-10-03"
215
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ subset:
226
+ fill: auto
227
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
228
+ checked_at: "2026-10-03"
229
+ unit.pointer_count:
230
+ fill: auto
231
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
232
+ checked_at: "2026-10-03"
233
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
235
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/debian-ubuntu-binary-packages.yaml ADDED
@@ -0,0 +1,265 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: debian-ubuntu-binary-packages
8
+ parent: null
9
+ title: "Debian and Ubuntu binary packages"
10
+ description: "Binary packages (.deb) listed in the official Debian and Ubuntu archives (Debian stable, oldstable, testing; Ubuntu resolute, noble, each with its -updates and -security suites). The package lists, sizes and hashes come from the Packages.xz index of each suite, component and architecture, as listed in the suite's InRelease file; one pointer per .deb file, and one per Packages.xz index. Packages can contain malicious code; treat files as data and run nothing outside a sandbox."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per package (see /usr/share/doc/<package>/copyright inside each package)"
20
+ licence_url: "https://www.debian.org/social_contract#guidelines"
21
+ licence_statement: "The Packages index has no license field. Debian main and contrib and Ubuntu main and universe hold free software meeting the Debian Free Software Guidelines; Debian non-free and non-free-firmware and Ubuntu restricted and multiverse hold software under per-package terms that are not free."
22
+ per_collection: null
23
+ training_ok: conditional
24
+ redistribution_ok: conditional
25
+ attribution: null
26
+ per_file: "Every row carries license_override describing its archive area."
27
+ licence_tiers: {unknown/unknown: 48503, yes/conditional: 627727}
28
+ spdx_top: {"(none)": 676230}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 676230
34
+ units: {file: 676230}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 2110641029932
38
+ rows_without_size: 0
39
+ collections:
40
+ distinct: 124090
41
+ top:
42
+ libunbound-dev: 27
43
+ libunbound8: 27
44
+ linux-cloud-tools-common: 29
45
+ linux-doc: 37
46
+ linux-tools-common: 29
47
+ linux-tools-host: 29
48
+ packages-index: 192
49
+ python3-unbound: 27
50
+ snapd: 27
51
+ unbound: 27
52
+ sub_collections: null
53
+ batches: null
54
+ n_ext: 2
55
+ usable_unit: null
56
+ form:
57
+ by_ext: {deb: 676038, xz: 192}
58
+ bytes_p50: 71996
59
+ bytes_p95: 9707706
60
+ bytes_max: 3341147280
61
+ size_quantiles:
62
+ max: 3341147280
63
+ min: 32
64
+ p25: 18626
65
+ p50: 71996
66
+ p75: 456368
67
+ p90: 3674792
68
+ p95: 9707706
69
+ p99: 92337506
70
+ p99.9: 224725856
71
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
72
+ by_ext_size:
73
+ - {ext: deb, rows: 676038, bytes: 2110444658528, p50: 72020, p95: 9708904, max: 3341147280}
74
+ - {ext: xz, rows: 192, bytes: 196371404, p50: 6936, p95: 9436656, max: 16202372}
75
+ over_cap:
76
+ cap: 61944279
77
+ rows_over: 8718
78
+ bytes_over: 1153717805672
79
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
80
+ largest:
81
+ - pointer_id: "debian-ubuntu-binary-packages:qgis-api-doc:debian/pool/main/q/qgis/qgis-api-doc_3.44.14+dfsg-1_all.deb"
82
+ size_bytes: 3341147280
83
+ - pointer_id: "debian-ubuntu-binary-packages:qgis-api-doc:ubuntu/pool/universe/q/qgis/qgis-api-doc_3.40.15+dfsg-1_all.deb"
84
+ size_bytes: 3197044082
85
+ - pointer_id: "debian-ubuntu-binary-packages:qgis-api-doc:debian/pool/main/q/qgis/qgis-api-doc_3.40.6+dfsg-1_all.deb"
86
+ size_bytes: 3179722260
87
+ - pointer_id: "debian-ubuntu-binary-packages:qgis-api-doc:ubuntu/pool/universe/q/qgis/qgis-api-doc_3.34.4+dfsg-4build2_all.deb"
88
+ size_bytes: 2893772000
89
+ - pointer_id: "debian-ubuntu-binary-packages:ceph-test-dbg:debian/pool/main/c/ceph/ceph-test-dbg_20.2.4+ds-1_amd64.deb"
90
+ size_bytes: 1910276688
91
+ date_years: null
92
+ media_type: {"(none)": 676230}
93
+ facets:
94
+ architecture: {all: 154922, amd64: 208429, arm64: 214902, i386: 97977}
95
+ component:
96
+ contrib: 1591
97
+ main: 422551
98
+ multiverse: 4250
99
+ non-free: 3092
100
+ non-free-firmware: 204
101
+ restricted: 40957
102
+ universe: 203585
103
+ distro: {debian: 386719, ubuntu: 289511}
104
+ priority: {extra: 27522, important: 1031, optional: 645855, required: 740, standard: 890}
105
+ release_date:
106
+ "Sat, 03 Oct 2026 8:50:11 UTC": 8
107
+ "Sat, 03 Oct 2026 8:50:27 UTC": 4
108
+ "Sat, 03 Oct 2026 8:51:07 UTC": 8
109
+ "Sat, 03 Oct 2026 8:51:46 UTC": 4
110
+ "Sat, 03 Oct 2026 08:09:36 UTC": 64
111
+ "Sat, 03 Oct 2026 08:45:53 UTC": 48
112
+ "Sat, 03 Oct 2026 13:03:47 UTC": 4
113
+ "Sat, 03 Oct 2026 13:04:14 UTC": 4
114
+ "Sat, 11 Jul 2026 10:16:37 UTC": 16
115
+ "Sat, 12 Sep 2026 07:55:41 UTC": 16
116
+ "Thu, 23 Apr 2026 17:07:15 UTC": 8
117
+ "Thu, 25 Apr 2024 15:10:33 UTC": 8
118
+ suite:
119
+ bookworm: 115981
120
+ bookworm-security: 6203
121
+ bookworm-updates: 115
122
+ forky: 121566
123
+ forky-security: 16
124
+ forky-updates: 16
125
+ noble: 100729
126
+ noble-security: 7999
127
+ noble-updates: 49948
128
+ resolute: 113471
129
+ resolute-security: 2843
130
+ resolute-updates: 14521
131
+ trixie: 138444
132
+ trixie-security: 4316
133
+ trixie-updates: 62
134
+ artifact: null
135
+ source_origin: null
136
+ parse_cost: null
137
+ parse_quirks: null
138
+ structure:
139
+ self_describing: null
140
+ integrity_builtin: null
141
+ has_reference_output: null
142
+ invariants: null
143
+ tools: null
144
+ ships:
145
+ problem_statement: {present: null}
146
+ reference_solution: {share: null}
147
+ tests: {present: null}
148
+ env_definition: {present: null}
149
+ known_answer: {present: null}
150
+ generator: {present: null}
151
+ upstream_labels: {present: null}
152
+ class: null
153
+ subject:
154
+ domain_span: null
155
+ shape_span: null
156
+ skill_family_span: null
157
+ ont_domain: null
158
+ tech_vocabulary: null
159
+ risk:
160
+ contamination: null
161
+ overlap_check: null
162
+ pii: null
163
+ pii_evidence: null
164
+ malware: null
165
+ ship_bytes: conditional
166
+ subset:
167
+ rule: "take all rows; licence, size and version are labels, not filters"
168
+ rows_in_subset: 676230
169
+ last_built: null
170
+ access:
171
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
172
+ verify: "Compare size_bytes and sha256 (Size and SHA256 in Packages; a few old entries without SHA256 carry md5 from MD5sum instead). Index rows use the size and sha256 published in the suite's Release."
173
+ hosts:
174
+ "http://archive.ubuntu.com": 178765
175
+ "http://ports.ubuntu.com": 110746
176
+ "https://deb.debian.org": 376184
177
+ "https://security.debian.org": 10535
178
+ uri_example: "https://deb.debian.org/debian/pool/main/0/0ad-data/0ad-data-common_0.0.26-1_all.deb"
179
+ needs_auth: {"false": 676230}
180
+ credential: null
181
+ hash_source: {source_published: 676230}
182
+ range_supported: null
183
+ reachable:
184
+ index_flag: {unknown: 676230}
185
+ index_check: null
186
+ our_sample: null
187
+ drift: null
188
+ sample: null
189
+ evidence:
190
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
191
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
192
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
193
+ class: {fill: unfilled, basis: "ships not judged"}
194
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
195
+ form.by_ext:
196
+ fill: auto
197
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
198
+ checked_at: "2026-10-03"
199
+ form.date_years:
200
+ fill: auto
201
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
202
+ checked_at: "2026-10-03"
203
+ form.over_cap:
204
+ fill: auto
205
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
206
+ checked_at: "2026-10-03"
207
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ form.size_quantiles:
210
+ fill: auto
211
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
212
+ checked_at: "2026-10-03"
213
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
214
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
215
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ provenance.licence_tiers:
217
+ fill: auto
218
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
219
+ checked_at: "2026-10-03"
220
+ provenance.spdx_top:
221
+ fill: auto
222
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
223
+ checked_at: "2026-10-03"
224
+ risk.contamination: {fill: unfilled, basis: "this source was not part of the benchmark overlap scan"}
225
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
228
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
229
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
237
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
238
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
239
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
240
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
241
+ snapshot.rev:
242
+ fill: auto
243
+ basis: "last commit listed in the publish receipt of this source"
244
+ checked_at: "2026-10-03"
245
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
247
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
249
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
250
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
251
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
252
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
253
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ subset:
256
+ fill: auto
257
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
258
+ checked_at: "2026-10-03"
259
+ unit.pointer_count:
260
+ fill: auto
261
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
262
+ checked_at: "2026-10-03"
263
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
265
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/discourse-forums.yaml ADDED
@@ -0,0 +1,284 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: discourse-forums
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: "259c2fc0884835fc284a4672d2327702d482ff19", at: "2026-10-03", rev_candidates: 2}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: null
17
+ licence_url: null
18
+ licence_statement: null
19
+ per_collection: null
20
+ training_ok: unknown
21
+ redistribution_ok: unknown
22
+ attribution: null
23
+ per_file: null
24
+ licence_tiers:
25
+ conditional/conditional: 2642937
26
+ no/unknown: 17795
27
+ unknown/unknown: 4390461
28
+ yes/conditional: 795546
29
+ yes/yes: 308811
30
+ spdx_top:
31
+ "Apache-2.0 WITH Swift-exception": 308811
32
+ CC-BY-NC-SA-3.0: 2642937
33
+ CC-BY-SA-4.0: 795546
34
+ "Per forum: CC BY-NC-SA 3.0, CC BY-SA 4.0, the Swift license, or none stated": 4408256
35
+ spdx_distinct: 4
36
+ licence_sample_check: null
37
+ unit:
38
+ pointer_level: null
39
+ pointer_count: 8155550
40
+ units: {file: 8155550}
41
+ state: null
42
+ expanded_count: null
43
+ bytes_total: 6178008382
44
+ rows_without_size: 0
45
+ collections:
46
+ distinct: 35
47
+ top:
48
+ community.home-assistant.io: 740423
49
+ community.letsencrypt.org: 338907
50
+ discourse.julialang.org: 523106
51
+ discuss.elastic.co: 740424
52
+ discussion.fedoraproject.org: 365148
53
+ forum.arduino.cc: 740330
54
+ forum.endeavouros.com: 509623
55
+ forum.manjaro.org: 498333
56
+ forums.swift.org: 308811
57
+ users.rust-lang.org: 373384
58
+ sub_collections: null
59
+ batches: null
60
+ n_ext: 1
61
+ usable_unit: null
62
+ form:
63
+ by_ext: {md: 8155550}
64
+ bytes_p50: 316
65
+ bytes_p95: 2432
66
+ bytes_max: 986282
67
+ size_quantiles:
68
+ max: 986282
69
+ min: 0
70
+ p25: 146
71
+ p50: 316
72
+ p75: 689
73
+ p90: 1472
74
+ p95: 2432
75
+ p99: 7382
76
+ p99.9: 28436
77
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
78
+ by_ext_size:
79
+ - {ext: md, rows: 8155550, bytes: 6178008382, p50: 316, p95: 2432, max: 986282}
80
+ over_cap:
81
+ cap: 61944279
82
+ rows_over: 0
83
+ bytes_over: 0
84
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
85
+ largest:
86
+ - {pointer_id: "discourse-forums:forums.swift.org:t6445/p6448@v1", size_bytes: 986282}
87
+ - {pointer_id: "discourse-forums:forums.swift.org:t6443/p6446@v1", size_bytes: 986277}
88
+ - {pointer_id: "discourse-forums:forums.swift.org:t6444/p6447@v1", size_bytes: 986203}
89
+ - {pointer_id: "discourse-forums:forums.swift.org:t6443/p19232@v1", size_bytes: 966476}
90
+ - {pointer_id: "discourse-forums:forums.swift.org:t6443/p19231@v1", size_bytes: 942914}
91
+ date_years:
92
+ key: created_at
93
+ rows_with_date: 8155550
94
+ counts:
95
+ "2011": 1
96
+ "2014": 17548
97
+ "2015": 75710
98
+ "2016": 172589
99
+ "2017": 306630
100
+ "2018": 432141
101
+ "2019": 504765
102
+ "2020": 770018
103
+ "2021": 816962
104
+ "2022": 798939
105
+ "2023": 880684
106
+ "2024": 1488747
107
+ "2025": 1262132
108
+ "2026": 628684
109
+ media_type: {text/markdown: 8155550}
110
+ facets:
111
+ instance:
112
+ community.deeplearning.ai: 199951
113
+ community.grafana.com: 146654
114
+ community.home-assistant.io: 740423
115
+ community.letsencrypt.org: 338907
116
+ community.n8n.io: 232037
117
+ discourse.julialang.org: 523106
118
+ discourse.nixos.org: 183744
119
+ discourse.pi-hole.net: 171727
120
+ discuss.elastic.co: 740424
121
+ discuss.python.org: 177611
122
+ discuss.pytorch.org: 282463
123
+ discussion.fedoraproject.org: 365148
124
+ forum.arduino.cc: 740330
125
+ forum.cursor.com: 189637
126
+ forum.endeavouros.com: 509623
127
+ forum.manjaro.org: 498333
128
+ forum.obsidian.md: 251129
129
+ forums.fast.ai: 170285
130
+ forums.swift.org: 308811
131
+ users.rust-lang.org: 373384
132
+ artifact: null
133
+ source_origin: null
134
+ parse_cost: null
135
+ parse_quirks: null
136
+ structure:
137
+ self_describing: null
138
+ integrity_builtin: null
139
+ has_reference_output: null
140
+ invariants: null
141
+ tools: null
142
+ ships:
143
+ problem_statement: {present: null}
144
+ reference_solution: {share: null}
145
+ tests: {present: null}
146
+ env_definition: {present: null}
147
+ known_answer: {present: null}
148
+ generator: {present: null}
149
+ upstream_labels: {present: null}
150
+ class: null
151
+ subject:
152
+ domain_span: null
153
+ shape_span: null
154
+ skill_family_span: null
155
+ ont_domain: null
156
+ tech_vocabulary: null
157
+ risk:
158
+ contamination: none
159
+ overlap_check:
160
+ done: true
161
+ pointers_flagged: 1
162
+ pointers_removed: 0
163
+ by_class: {trivial_small: 1}
164
+ benches: {NL2Repo: 1}
165
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
166
+ pii: null
167
+ pii_evidence: null
168
+ malware: null
169
+ ship_bytes: unknown
170
+ subset:
171
+ rule: "take all rows; licence, size and version are labels, not filters"
172
+ rows_in_subset: 8155550
173
+ last_built:
174
+ date: "2026-10-02"
175
+ rule: "older rule that filtered by licence tier"
176
+ rows: 5167761
177
+ rows_in_subset: 2427907
178
+ excluded_by_licence: 2724219
179
+ duplicates: 49506
180
+ access:
181
+ method: null
182
+ verify: null
183
+ hosts:
184
+ "https://community.home-assistant.io": 740423
185
+ "https://community.letsencrypt.org": 338907
186
+ "https://discourse.julialang.org": 523106
187
+ "https://discuss.elastic.co": 740424
188
+ "https://discussion.fedoraproject.org": 365148
189
+ "https://forum.arduino.cc": 740330
190
+ "https://forum.endeavouros.com": 509623
191
+ "https://forum.manjaro.org": 498333
192
+ "https://forums.swift.org": 308811
193
+ "https://users.rust-lang.org": 373384
194
+ uri_example: "https://community.deeplearning.ai/posts/164711.json"
195
+ needs_auth: {"false": 8155550}
196
+ credential: null
197
+ hash_source: {computed: 8155550}
198
+ range_supported: null
199
+ reachable:
200
+ index_flag: {"(none)": 8155550}
201
+ index_check: null
202
+ our_sample: null
203
+ drift: null
204
+ sample: null
205
+ evidence:
206
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ class: {fill: unfilled, basis: "ships not judged"}
210
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ form.by_ext:
212
+ fill: auto
213
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
214
+ checked_at: "2026-10-03"
215
+ form.date_years:
216
+ fill: auto
217
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
218
+ checked_at: "2026-10-03"
219
+ form.over_cap:
220
+ fill: auto
221
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
222
+ checked_at: "2026-10-03"
223
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ form.size_quantiles:
226
+ fill: auto
227
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
228
+ checked_at: "2026-10-03"
229
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
231
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ provenance.licence_tiers:
233
+ fill: auto
234
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
235
+ checked_at: "2026-10-03"
236
+ provenance.spdx_top:
237
+ fill: auto
238
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
239
+ checked_at: "2026-10-03"
240
+ risk.contamination:
241
+ fill: auto
242
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
243
+ checked_at: "2026-10-03"
244
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
245
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
247
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
248
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
249
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
250
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
251
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
252
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
253
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
257
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
260
+ snapshot.rev:
261
+ fill: auto
262
+ basis: "last commit listed in the publish receipt of this source"
263
+ checked_at: "2026-10-03"
264
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
265
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
266
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
268
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
270
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
272
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
273
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ subset:
275
+ fill: auto
276
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
277
+ checked_at: "2026-10-03"
278
+ unit.pointer_count:
279
+ fill: auto
280
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
281
+ checked_at: "2026-10-03"
282
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
284
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/docker-hub.yaml ADDED
@@ -0,0 +1,268 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: docker-hub
8
+ parent: null
9
+ title: "Docker Hub container images"
10
+ description: "Container images on Docker Hub, listed through the Docker Hub API: repositories of the Docker Official Images (namespace library); a hand-picked list of Verified Publisher and Sponsored OSS namespaces. One pointer per platform image of a tag (for example linux/amd64); the uri is the image manifest in the Docker registry, addressed by digest, so the sha256 is the digest itself. size_bytes is the compressed size of the image layers as reported by Docker Hub. Repository and tag details are in source_meta. Image layers are not listed and not stored."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per-image licenses; Docker Terms of Service"
20
+ licence_url: "https://www.docker.com/legal/docker-terms-service/"
21
+ licence_statement: "Docker Hub gives no license field. Each image bundles software under its own licenses; the terms of service say: \"All third party images shall be deemed Third-Party Content subject to their corresponding terms and conditions.\" Check each image before use. The terms restrict automated access and training of competing AI models (quoted in the notes)."
22
+ per_collection: null
23
+ training_ok: conditional
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {conditional/unknown: 593368}
28
+ spdx_top: {"(none)": 593368}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 593368
34
+ units: {file: 593368}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 221139052776294
38
+ rows_without_size: 0
39
+ collections:
40
+ distinct: 30
41
+ top:
42
+ amazon: 19987
43
+ circleci: 27241
44
+ confluentinc: 31523
45
+ datadog: 31855
46
+ docker: 31209
47
+ grafana: 56340
48
+ hashicorp: 16222
49
+ library: 142734
50
+ linuxserver: 89518
51
+ newrelic: 19544
52
+ sub_collections: null
53
+ batches: null
54
+ n_ext: 1
55
+ usable_unit: null
56
+ form:
57
+ by_ext: {json: 593368}
58
+ bytes_p50: 131740034
59
+ bytes_p95: 1465391198
60
+ bytes_max: 23867383457
61
+ size_quantiles:
62
+ max: 23867383457
63
+ min: 0
64
+ p25: 45211487
65
+ p50: 131740034
66
+ p75: 357516254
67
+ p90: 1005505542
68
+ p95: 1465391198
69
+ p99: 2953556688
70
+ p99.9: 14712399019
71
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
72
+ by_ext_size:
73
+ - {ext: json, rows: 593368, bytes: 221139052776294, p50: 131740034, p95: 1465391198, max: 23867383457}
74
+ over_cap:
75
+ cap: 61944279
76
+ rows_over: 399099
77
+ bytes_over: 215746984333155
78
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
79
+ largest:
80
+ - pointer_id: "docker-hub:elastic:package-registry-distribution/8.19.22-ubi/linux-amd64@c4e43469f4d88d9ed980613e3feff4f2bb1ee6155b1876d9f7513d48b4f48234"
81
+ size_bytes: 23867383457
82
+ - pointer_id: "docker-hub:elastic:package-registry-distribution/8.19.22-ubi/linux-arm64@3b7089cb47e1d6e527597cfb6fd20f89b45877a735f8459505fe1c91c3d99e6a"
83
+ size_bytes: 23864336576
84
+ - pointer_id: "docker-hub:elastic:package-registry-distribution/8.19.22-fips/linux-amd64@5fe7e16e45d46e7fd8a5551a39bf2312d6b5a986a58aec124ba25469b1b9f508"
85
+ size_bytes: 23842966945
86
+ - pointer_id: "docker-hub:elastic:package-registry-distribution/8.19.22/linux-amd64@d46ca648584c7ef315cb372a84118bd6a76dc39db580782ad6d720942819b37d"
87
+ size_bytes: 23842058294
88
+ - pointer_id: "docker-hub:elastic:package-registry-distribution/8.19.22-fips/linux-arm64@31997eca19bdfdd03c674afab2131ddbdbf4a32bd1dbcaf434ba0769cb300665"
89
+ size_bytes: 23840248854
90
+ date_years: null
91
+ media_type: {"(none)": 593368}
92
+ facets:
93
+ arch:
94
+ "": 724
95
+ "386": 15209
96
+ amd64: 276152
97
+ arm: 53557
98
+ arm64: 182644
99
+ mips: 309
100
+ mips64: 309
101
+ mips64le: 2576
102
+ mipsle: 309
103
+ ppc64le: 27980
104
+ riscv64: 7757
105
+ s390x: 25840
106
+ x86_64: 2
107
+ hub_tier: {official: 142734, publisher_list: 450634}
108
+ is_official: {"false": 450634, "true": 142734}
109
+ namespace:
110
+ amazon: 19987
111
+ apache: 9652
112
+ cimg: 7818
113
+ circleci: 27241
114
+ confluentinc: 31523
115
+ datadog: 31855
116
+ docker: 31209
117
+ elastic: 11591
118
+ gitlab: 11585
119
+ grafana: 56340
120
+ hashicorp: 16222
121
+ jenkins: 8494
122
+ jetbrains: 6414
123
+ library: 142734
124
+ linuxserver: 89518
125
+ mongodb: 11512
126
+ newrelic: 19544
127
+ nginxinc: 7291
128
+ portainer: 9929
129
+ ubuntu: 10712
130
+ os: {"": 1420, darwin: 2700, freebsd: 186, linux: 566991, netbsd: 144, openbsd: 186, windows: 21741}
131
+ variant: {v5: 4223, v6: 8772, v7: 34630, v8: 50157}
132
+ artifact: null
133
+ source_origin: null
134
+ parse_cost: null
135
+ parse_quirks: null
136
+ structure:
137
+ self_describing: null
138
+ integrity_builtin: null
139
+ has_reference_output: null
140
+ invariants: null
141
+ tools: null
142
+ ships:
143
+ problem_statement: {present: null}
144
+ reference_solution: {share: null}
145
+ tests: {present: null}
146
+ env_definition: {present: null}
147
+ known_answer: {present: null}
148
+ generator: {present: null}
149
+ upstream_labels: {present: null}
150
+ class: null
151
+ subject:
152
+ domain_span: null
153
+ shape_span: null
154
+ skill_family_span: null
155
+ ont_domain: null
156
+ tech_vocabulary: null
157
+ risk:
158
+ contamination: none
159
+ overlap_check:
160
+ done: true
161
+ pointers_flagged: 38
162
+ pointers_removed: 0
163
+ by_class: {base_image: 38}
164
+ benches: {"FrontierSWE v2": 10, Terminal-Bench: 33}
165
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
166
+ pii: null
167
+ pii_evidence: null
168
+ malware: null
169
+ ship_bytes: unknown
170
+ subset:
171
+ rule: "take all rows; licence, size and version are labels, not filters"
172
+ rows_in_subset: 593368
173
+ last_built: null
174
+ access:
175
+ method: "GET a pull token from https://auth.docker.io/token?service=registry.docker.io&scope=repository:<namespace>/<repo>:pull (anonymous, no account), then GET uri with the header Authorization: Bearer <token> and Accept: application/vnd.oci.image.manifest.v1+json, application/vnd.docker.distribution.manifest.v2+json, application/vnd.oci.image.index.v1+json, application/vnd.docker.distribution.manifest.list.v2+json. Every manifest GET counts as a pull: anonymous users get 100 pulls per 6 hours per IPv4 address (or IPv6 /64). The Docker Hub API allows 180 requests per minute; this listing made at most one request per second."
176
+ verify: "Compare the sha256 of the manifest bytes as received with hashes.sha256 (it equals the digest in the uri). size_bytes is not the manifest length: it is the sum of layers[].size inside the manifest, so compare that sum instead. The stored file is the manifest."
177
+ hosts: {"https://registry-1.docker.io": 593368}
178
+ uri_example: "https://registry-1.docker.io/v2/amazon/amazon-ec2-instance-selector/manifests/sha256:fab522ba525fc7d41689bc60b99c3ff4e1ef0d21fe571638e9d04263e8af9f25"
179
+ needs_auth: {"false": 593368}
180
+ credential: null
181
+ hash_source: {source_published: 593368}
182
+ range_supported: null
183
+ reachable:
184
+ index_flag: {unknown: 593368}
185
+ index_check: null
186
+ our_sample: null
187
+ drift: null
188
+ sample: null
189
+ evidence:
190
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
191
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
192
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
193
+ class: {fill: unfilled, basis: "ships not judged"}
194
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
195
+ form.by_ext:
196
+ fill: auto
197
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
198
+ checked_at: "2026-10-03"
199
+ form.date_years:
200
+ fill: auto
201
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
202
+ checked_at: "2026-10-03"
203
+ form.over_cap:
204
+ fill: auto
205
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
206
+ checked_at: "2026-10-03"
207
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ form.size_quantiles:
210
+ fill: auto
211
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
212
+ checked_at: "2026-10-03"
213
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
214
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
215
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ provenance.licence_tiers:
217
+ fill: auto
218
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
219
+ checked_at: "2026-10-03"
220
+ provenance.spdx_top:
221
+ fill: auto
222
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
223
+ checked_at: "2026-10-03"
224
+ risk.contamination:
225
+ fill: auto
226
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
227
+ checked_at: "2026-10-03"
228
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
232
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
237
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
238
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
239
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
240
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
241
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
242
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
243
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
244
+ snapshot.rev:
245
+ fill: auto
246
+ basis: "last commit listed in the publish receipt of this source"
247
+ checked_at: "2026-10-03"
248
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
249
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
250
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
251
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
252
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
253
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
257
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ subset:
259
+ fill: auto
260
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
261
+ checked_at: "2026-10-03"
262
+ unit.pointer_count:
263
+ fill: auto
264
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
265
+ checked_at: "2026-10-03"
266
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
268
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/egov-japan-data.yaml ADDED
@@ -0,0 +1,373 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: egov-japan-data
8
+ parent: null
9
+ title: "e-Gov Data Portal (Japan)"
10
+ description: "Data files (CSV, spreadsheets, PDF, JSON, geodata and more) registered on the e-Gov Data Portal, the Japanese national government open data portal run by the Digital Agency. Datasets were listed through the portal's CKAN Action API (package_search, sorted by modification time). One pointer per resource; most files sit on the ministries' and agencies' own servers, a few on the portal's own storage."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per resource (Government of Japan Standard Terms of Use 2.0, CC BY 4.0, Public Data License 1.0, CC0 and others)"
20
+ licence_url: "https://data.e-gov.go.jp/info/ja/terms"
21
+ licence_statement: "Each resource carries its own license_id and license_title, set by the publishing organization. Most are the Government of Japan Standard Terms of Use (version 2.0) or CC BY, which allow reuse including commercial use with attribution; a few resources give no license, a non-commercial or no-derivatives license, or note third-party rights. The portal's own terms say that, unless stated otherwise, content on the site is under the Public Data License (version 1.0) and the source must be credited."
22
+ per_collection: null
23
+ training_ok: conditional
24
+ redistribution_ok: conditional
25
+ attribution: "Credit the source as the terms ask, for example the publishing organization and the dataset title with its URL (source_meta.organization, title, dataset_url). The Standard Terms of Use and the Public Data License ask that edited material be marked as edited by the user."
26
+ per_file: "license_override gives each row's license (license_id, license_title and any copyright note, from the resource or, when the resource has none, from the dataset) with its own training_ok and redistribution_ok; source_meta repeats the raw values."
27
+ licence_tiers: {conditional/conditional: 3, unknown/unknown: 2764, yes/yes: 289283}
28
+ spdx_top: {"(none)": 292041, CC-BY-2.1-JP: 1, CC-BY-4.0: 8}
29
+ spdx_distinct: 3
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 292050
34
+ units: {file: 292050}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 15444483373
38
+ rows_without_size: 244867
39
+ collections:
40
+ distinct: 23
41
+ top:
42
+ org_0400: 32694
43
+ org_0600: 6102
44
+ org_1100: 21577
45
+ org_1200: 20509
46
+ org_1400: 5919
47
+ org_1500: 17379
48
+ org_1600: 41181
49
+ org_1700: 104656
50
+ org_1900: 6060
51
+ org_2000: 12404
52
+ sub_collections: null
53
+ batches: null
54
+ n_ext: 35
55
+ usable_unit: null
56
+ form:
57
+ by_ext:
58
+ html: 69028
59
+ do: 66833
60
+ xls: 62223
61
+ pdf: 41859
62
+ csv: 28412
63
+ htm: 10806
64
+ xlsx: 8406
65
+ gif: 1659
66
+ zip: 1266
67
+ xml: 652
68
+ jpg: 223
69
+ kmz: 201
70
+ "(none)": 112
71
+ epub: 84
72
+ mp3: 56
73
+ php: 41
74
+ docx: 30
75
+ doc: 29
76
+ asp: 27
77
+ pptx: 22
78
+ "(15 other types)": 81
79
+ bytes_p50: 47696
80
+ bytes_p95: 837667
81
+ bytes_max: 144211603
82
+ size_quantiles:
83
+ max: 144211603
84
+ min: 66
85
+ p25: 22820
86
+ p50: 47696
87
+ p75: 120320
88
+ p90: 364933
89
+ p95: 837667
90
+ p99: 4704123
91
+ p99.9: 37531730
92
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
93
+ by_ext_size:
94
+ - {ext: html, rows: 69028, bytes: 316872593, p50: 30337, p95: 122000, max: 1163109}
95
+ - {ext: do, rows: 66833, bytes: 2486590138, p50: 88576, p95: 666624, max: 63937024}
96
+ - {ext: xls, rows: 62223, bytes: 1023305311, p50: 44144, p95: 189500, max: 21827584}
97
+ - {ext: pdf, rows: 41859, bytes: 10416498368, p50: 392914, p95: 6400463, max: 144211603}
98
+ - {ext: csv, rows: 28412, bytes: 154917869, p50: 4737, p95: 204997, max: 1196521}
99
+ - {ext: htm, rows: 10806, bytes: 18972486, p50: 11651, p95: 69866, max: 164955}
100
+ - {ext: xlsx, rows: 8406, bytes: 370720113, p50: 21016, p95: 395985, max: 31302539}
101
+ - {ext: gif, rows: 1659, bytes: 18539785, p50: 48385, p95: 104310, max: 245242}
102
+ - {ext: zip, rows: 1266, bytes: 419999047, p50: 462709, p95: 6086883, max: 63130494}
103
+ - {ext: xml, rows: 652, bytes: 5078452, p50: 9710, p95: 13038, max: 3223687}
104
+ - {ext: jpg, rows: 223, bytes: 180794947, p50: 4429995, p95: 5759077, max: 6763045}
105
+ - {ext: kmz, rows: 201, bytes: 6030803, p50: 47999, p95: 396544, max: 524967}
106
+ - {ext: "(none)", rows: 112, bytes: 17348, p50: 17348, p95: 17348, max: 17348}
107
+ - {ext: epub, rows: 84, bytes: 23217923, p50: 576069, p95: 10504881, max: 10504881}
108
+ - {ext: mp3, rows: 56, bytes: 0, p50: null, p95: null, max: null}
109
+ - {ext: php, rows: 41, bytes: 0, p50: null, p95: null, max: null}
110
+ - {ext: docx, rows: 30, bytes: 121853, p50: 59592, p95: 62261, max: 62261}
111
+ - {ext: doc, rows: 29, bytes: 464568, p50: 96952, p95: 314368, max: 314368}
112
+ - {ext: asp, rows: 27, bytes: 0, p50: null, p95: null, max: null}
113
+ - {ext: pptx, rows: 22, bytes: 2287578, p50: 488596, p95: 736780, max: 736780}
114
+ over_cap:
115
+ cap: 61944279
116
+ rows_over: 29
117
+ bytes_over: 2560290305
118
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
119
+ largest:
120
+ - pointer_id: "egov-japan-data:org_0800:8e165887-5853-41c9-9dc1-5fcba4353b2d/h1.pdf"
121
+ size_bytes: 144211603
122
+ - pointer_id: "egov-japan-data:org_1600:1d62a851-7ba4-4d8d-b3eb-369716391dd5/file-download"
123
+ size_bytes: 121054541
124
+ - pointer_id: "egov-japan-data:org_1900:b96be088-36b8-47bf-9797-d000e2200e17/523546_DS1.pdf"
125
+ size_bytes: 120270433
126
+ - pointer_id: "egov-japan-data:org_1600:3af280e2-c5cb-4eb9-a030-fbe22fae35ee/file-download"
127
+ size_bytes: 118365410
128
+ - pointer_id: "egov-japan-data:org_1600:4be8d13e-1749-4faf-888a-7bb04c0281e4/file-download"
129
+ size_bytes: 116862368
130
+ date_years: null
131
+ media_type:
132
+ "(none)": 288004
133
+ application/pdf: 2573
134
+ application/vnd.ms-excel: 40
135
+ application/vnd.openxmlformats-officedocument.presentationml.presentation: 3
136
+ application/vnd.openxmlformats-officedocument.spreadsheetml.sheet: 592
137
+ application/vnd.openxmlformats-officedocument.wordprocessingml.document: 4
138
+ application/zip: 27
139
+ text/csv: 35
140
+ text/html: 754
141
+ video/mp4: 13
142
+ facets:
143
+ format:
144
+ " ": 493
145
+ CSV: 28856
146
+ DOC: 30
147
+ DOCX: 30
148
+ EXE: 71
149
+ GIF: 1659
150
+ HTML: 78687
151
+ JPEG: 229
152
+ KMZ: 201
153
+ MP3: 56
154
+ PDF: 43778
155
+ PNG: 23
156
+ PPTX: 22
157
+ TXT: 18
158
+ XLS: 121834
159
+ XLSX: 12890
160
+ XML: 1769
161
+ ZIP: 1265
162
+ epub: 86
163
+ head_content_type:
164
+ application/epub+zip: 1
165
+ application/msword: 3
166
+ application/octet-stream: 24477
167
+ application/pdf: 6728
168
+ application/vnd.google-earth.kmz: 53
169
+ application/vnd.ms-excel: 1414
170
+ application/zip: 236
171
+ binary/octet-stream: 154
172
+ image/gif: 351
173
+ image/jpeg: 43
174
+ text/csv: 1089
175
+ "text/csv;charset=UTF-8": 4256
176
+ text/html: 8707
177
+ "text/html; charset=UTF-8": 976
178
+ "text/html; charset=utf-8": 254
179
+ "text/html;charset=UTF-8": 13
180
+ "text/html;charset=euc-JP": 2
181
+ "text/html;charset=utf-8": 166
182
+ text/plain: 2
183
+ text/xml: 133
184
+ head_status:
185
+ "200": 49841
186
+ "400": 223
187
+ "403": 10
188
+ "404": 1334
189
+ "500": 1
190
+ blocked_host: 6406
191
+ net_error: 63
192
+ not_sent_robots: 692
193
+ hosted_on_portal: {"no": 291926, "yes": 124}
194
+ license_id:
195
+ CC-BY: 4714
196
+ cc-by: 177204
197
+ cc-by-2.1-jp: 1
198
+ cc-by-4.0: 8
199
+ cc-by-nc: 1
200
+ cc-by-nc-sa: 2
201
+ gjstu: 106703
202
+ pdl-1.0: 39
203
+ license_level: {dataset: 2764, resource: 289286}
204
+ organization: {}
205
+ organization_name:
206
+ org_0200: 1429
207
+ org_0300: 4079
208
+ org_0400: 32694
209
+ org_0500: 699
210
+ org_0600: 6102
211
+ org_0700: 3859
212
+ org_0800: 2054
213
+ org_0900: 600
214
+ org_0950: 3861
215
+ org_1000: 483
216
+ org_1100: 21577
217
+ org_1200: 20509
218
+ org_1300: 5419
219
+ org_1400: 5919
220
+ org_1500: 17379
221
+ org_1600: 41181
222
+ org_1700: 104656
223
+ org_1900: 6060
224
+ org_2000: 12404
225
+ org_2100: 918
226
+ size_source: {http_head: 47183, none: 244867}
227
+ artifact: null
228
+ source_origin: null
229
+ parse_cost: null
230
+ parse_quirks: null
231
+ structure:
232
+ self_describing: null
233
+ integrity_builtin: null
234
+ has_reference_output: null
235
+ invariants: null
236
+ tools: null
237
+ ships:
238
+ problem_statement: {present: null}
239
+ reference_solution: {share: null}
240
+ tests: {present: null}
241
+ env_definition: {present: null}
242
+ known_answer: {present: null}
243
+ generator: {present: null}
244
+ upstream_labels: {present: null}
245
+ class: null
246
+ subject:
247
+ domain_span: null
248
+ shape_span: null
249
+ skill_family_span: null
250
+ ont_domain: null
251
+ tech_vocabulary: null
252
+ risk:
253
+ contamination: none
254
+ overlap_check:
255
+ done: true
256
+ pointers_flagged: 0
257
+ pointers_removed: 0
258
+ by_class: {}
259
+ benches: {}
260
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
261
+ pii: null
262
+ pii_evidence: null
263
+ malware: null
264
+ ship_bytes: conditional
265
+ subset:
266
+ rule: "take all rows; licence, size and version are labels, not filters"
267
+ rows_in_subset: 292050
268
+ last_built: null
269
+ access:
270
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
271
+ verify: "size_bytes is given only when an HTTP HEAD request at listing time returned a Content-Length; otherwise it is null and the size is measured at retrieval. Hashes are computed at retrieval (hash_source computed_on_fetch). The portal's own size field is kept as source_meta.portal_size; it is often a rounded estimate and is not used for verification."
272
+ hosts:
273
+ "http://www.aec.go.jp": 4846
274
+ "http://www.e-stat.go.jp": 47789
275
+ "https://hakusyo1.moj.go.jp": 17005
276
+ "https://www.e-stat.go.jp": 98743
277
+ "https://www.env.go.jp": 9966
278
+ "https://www.jftc.go.jp": 6101
279
+ "https://www.mhlw.go.jp": 12525
280
+ "https://www.mofa.go.jp": 5251
281
+ "https://www.soumu.go.jp": 20444
282
+ "https://www5.cao.go.jp": 13833
283
+ uri_example: "https://www.gyoukaku.go.jp/review/database/H26/database2014.xlsx"
284
+ needs_auth: {"false": 292050}
285
+ credential: null
286
+ hash_source: {computed_on_fetch: 292050}
287
+ range_supported: null
288
+ reachable:
289
+ index_flag: {"(none)": 292050}
290
+ index_check: null
291
+ our_sample: null
292
+ drift: null
293
+ sample: null
294
+ evidence:
295
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
296
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
297
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
298
+ class: {fill: unfilled, basis: "ships not judged"}
299
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
300
+ form.by_ext:
301
+ fill: auto
302
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
303
+ checked_at: "2026-10-03"
304
+ form.date_years:
305
+ fill: auto
306
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
307
+ checked_at: "2026-10-03"
308
+ form.over_cap:
309
+ fill: auto
310
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
311
+ checked_at: "2026-10-03"
312
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
313
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
314
+ form.size_quantiles:
315
+ fill: auto
316
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
317
+ checked_at: "2026-10-03"
318
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
319
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
320
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
321
+ provenance.licence_tiers:
322
+ fill: auto
323
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
324
+ checked_at: "2026-10-03"
325
+ provenance.spdx_top:
326
+ fill: auto
327
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
328
+ checked_at: "2026-10-03"
329
+ risk.contamination:
330
+ fill: auto
331
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
332
+ checked_at: "2026-10-03"
333
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
334
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
335
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
336
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
337
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
338
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
339
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
340
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
341
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
342
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
343
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
344
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
345
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
346
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
347
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
348
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
349
+ snapshot.rev:
350
+ fill: auto
351
+ basis: "last commit listed in the publish receipt of this source"
352
+ checked_at: "2026-10-03"
353
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
354
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
355
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
356
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
357
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
358
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
359
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
360
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
361
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
362
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
363
+ subset:
364
+ fill: auto
365
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
366
+ checked_at: "2026-10-03"
367
+ unit.pointer_count:
368
+ fill: auto
369
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
370
+ checked_at: "2026-10-03"
371
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
372
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
373
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/end-of-term-web-archive.yaml ADDED
@@ -0,0 +1,302 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: end-of-term-web-archive
8
+ parent: null
9
+ title: "End of Term Web Archive"
10
+ description: "Office and document files (PDF, Word, Excel, PowerPoint, CSV, plain text and similar) captured by the End of Term Web Archive, which crawls U.S. federal government websites at each change of administration (terms 2004 to 2024). Each pointer is one HTTP response record inside a WARC file, retrieved by byte range; the hash covers the HTTP response body. Payloads are deduplicated across terms by hash."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: "0fbcf8f07930180eb3dd6a9187d6fcf35202f0f0", at: "2026-10-02", rev_candidates: 2}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "No restrictions (Creative Commons Zero)"
17
+ licence_url: "https://registry.opendata.aws/eot-web-archive/"
18
+ licence_statement: "The dataset page states: \"There are no restrictions on the use, access, and/or download of data from the End of Term Web Archive Dataset.\" It is marked Creative Commons Zero. Third-party content inside archived pages keeps its own terms."
19
+ per_collection: null
20
+ training_ok: "yes"
21
+ redistribution_ok: "yes"
22
+ attribution: "The project asks users to cite the End of Term Web Archive."
23
+ per_file: null
24
+ licence_tiers: {yes/yes: 337638097}
25
+ spdx_top: {"(none)": 337638097}
26
+ spdx_distinct: 1
27
+ licence_sample_check: null
28
+ unit:
29
+ pointer_level: null
30
+ pointer_count: 337638097
31
+ units: {"WARC record": 337638097}
32
+ state: null
33
+ expanded_count: null
34
+ bytes_total: 149402879676720
35
+ rows_without_size: 0
36
+ collections:
37
+ distinct: 6
38
+ top:
39
+ EOT-2004: 4839487
40
+ EOT-2008: 7446377
41
+ EOT-2012: 7792749
42
+ EOT-2016: 61400114
43
+ EOT-2020: 27585323
44
+ EOT-2024: 228574047
45
+ sub_collections: null
46
+ batches: null
47
+ n_ext: 22
48
+ usable_unit: null
49
+ form:
50
+ by_ext:
51
+ xml: 157511214
52
+ pdf: 88829829
53
+ txt: 79541460
54
+ csv: 3194216
55
+ doc: 2042567
56
+ docx: 1862273
57
+ xls: 1533613
58
+ xlsx: 1445123
59
+ ics: 797662
60
+ rtf: 341713
61
+ ppt: 312656
62
+ pptx: 176456
63
+ eml: 35227
64
+ odt: 5037
65
+ ods: 4522
66
+ xlsm: 2804
67
+ odp: 837
68
+ mbox: 445
69
+ pptm: 233
70
+ odf: 125
71
+ "(2 other types)": 85
72
+ bytes_p50: 14270
73
+ bytes_p95: 1118781
74
+ bytes_max: 43409202807
75
+ size_quantiles:
76
+ max: 43409202807
77
+ min: 100
78
+ p25: 3136
79
+ p50: 14270
80
+ p75: 45740
81
+ p90: 359174
82
+ p95: 1118781
83
+ p99: 6590921
84
+ p99.9: 49671848
85
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
86
+ by_ext_size:
87
+ - {ext: xml, rows: 157511214, bytes: 3972301592115, p50: 13395, p95: 27868, max: 1993973550}
88
+ - {ext: pdf, rows: 88829829, bytes: 111613761415493, p50: 170216, p95: 4434179, max: 6819913801}
89
+ - {ext: txt, rows: 79541460, bytes: 27994771804230, p50: 2418, p95: 106260, max: 43409202807}
90
+ - {ext: csv, rows: 3194216, bytes: 916851131201, p50: 3827, p95: 87462, max: 13119472026}
91
+ - {ext: doc, rows: 2042567, bytes: 360541065139, p50: 13875, p95: 533450, max: 434238461}
92
+ - {ext: docx, rows: 1862273, bytes: 720043866076, p50: 28995, p95: 1654984, max: 458140809}
93
+ - {ext: xls, rows: 1533613, bytes: 130245556946, p50: 9130, p95: 252649, max: 156870375}
94
+ - {ext: xlsx, rows: 1445123, bytes: 994085648154, p50: 22416, p95: 1879243, max: 887447398}
95
+ - {ext: ics, rows: 797662, bytes: 4839809982, p50: 1350, p95: 23501, max: 1338074}
96
+ - {ext: rtf, rows: 341713, bytes: 12597414687, p50: 6774, p95: 66620, max: 84388505}
97
+ - {ext: ppt, rows: 312656, bytes: 1030664445723, p50: 762693, p95: 14163497, max: 441775950}
98
+ - {ext: pptx, rows: 176456, bytes: 1630060212176, p50: 1808857, p95: 33808923, max: 2084880347}
99
+ - {ext: eml, rows: 35227, bytes: 1554977966, p50: 1438, p95: 67800, max: 61008521}
100
+ - {ext: odt, rows: 5037, bytes: 649717383, p50: 21269, p95: 156120, max: 53643795}
101
+ - {ext: ods, rows: 4522, bytes: 726537471, p50: 15857, p95: 182840, max: 120874189}
102
+ - {ext: xlsm, rows: 2804, bytes: 11071623207, p50: 838379, p95: 12744863, max: 955058548}
103
+ - {ext: odp, rows: 837, bytes: 4319377467, p50: 863374, p95: 26883062, max: 110986648}
104
+ - {ext: mbox, rows: 445, bytes: 1692882306, p50: 9392, p95: 156506, max: 888921269}
105
+ - {ext: pptm, rows: 233, bytes: 2084550714, p50: 3125706, p95: 27587256, max: 435080628}
106
+ - {ext: odf, rows: 125, bytes: 5933472, p50: 20792, p95: 133773, max: 782134}
107
+ over_cap:
108
+ cap: 61944279
109
+ rows_over: 259117
110
+ bytes_over: 42131498639322
111
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
112
+ largest:
113
+ - pointer_id: "end-of-term-web-archive:EOT-2024:EOT24PRE-20241031160823-crawl912_EOT24PRE-20241031160823-28875.warc.gz@378740466"
114
+ size_bytes: 43409202807
115
+ - pointer_id: "end-of-term-web-archive:EOT-2024:EOT24PRE-20240930011705-crawl912_EOT24PRE-20240930011705-08722.warc.gz@975686363"
116
+ size_bytes: 28721117830
117
+ - pointer_id: "end-of-term-web-archive:EOT-2024:EOT24PRE-20240929221734-crawl912_EOT24PRE-20240929221734-08361.warc.gz@856865203"
118
+ size_bytes: 25605649852
119
+ - pointer_id: "end-of-term-web-archive:EOT-2024:archiveteam_usgovernment_20250202065908_b49ce642_usgovernment_20250202065908_b49ce642.1738361595.megawarc.warc.gz@1695004735"
120
+ size_bytes: 20418271499
121
+ - pointer_id: "end-of-term-web-archive:EOT-2024:archiveteam_usgovernment_20250207011821_16cf2baf_usgovernment_20250207011821_16cf2baf.1738361595.megawarc.warc.gz@1369651351"
122
+ size_bytes: 20173711482
123
+ date_years:
124
+ key: origin.captured_at
125
+ rows_with_date: 337638097
126
+ counts:
127
+ "2004": 4839275
128
+ "2005": 212
129
+ "2008": 4340018
130
+ "2009": 3106359
131
+ "2012": 4716278
132
+ "2013": 3076471
133
+ "2016": 14304269
134
+ "2017": 47095845
135
+ "2020": 12994592
136
+ "2021": 14590731
137
+ "2024": 17863472
138
+ "2025": 210710575
139
+ media_type:
140
+ application/csv: 609065
141
+ application/msword: 1998086
142
+ application/pdf: 88829560
143
+ application/vnd.ms-excel: 1533525
144
+ application/vnd.openxmlformats-officedocument.spreadsheetml.sheet: 1445123
145
+ application/vnd.openxmlformats-officedocument.wordprocessingml.document: 1862273
146
+ text/calendar: 797662
147
+ text/csv: 2585136
148
+ text/plain: 79541460
149
+ text/xml: 157505429
150
+ facets: null
151
+ artifact: null
152
+ source_origin: null
153
+ parse_cost: null
154
+ parse_quirks: null
155
+ structure:
156
+ self_describing: null
157
+ integrity_builtin: null
158
+ has_reference_output: null
159
+ invariants: null
160
+ tools: null
161
+ ships:
162
+ problem_statement: {present: null}
163
+ reference_solution: {share: null}
164
+ tests: {present: null}
165
+ env_definition: {present: null}
166
+ known_answer: {present: null}
167
+ generator: {present: null}
168
+ upstream_labels: {present: null}
169
+ class: null
170
+ subject:
171
+ domain_span: null
172
+ shape_span: null
173
+ skill_family_span: null
174
+ ont_domain: null
175
+ tech_vocabulary: null
176
+ risk:
177
+ contamination: public_benchmark
178
+ overlap_check:
179
+ done: true
180
+ pointers_flagged: 581
181
+ pointers_removed: 441
182
+ by_class: {data_address: 419, generic_text: 4, reference_or_namespace: 67, task_file: 22, trivial_small: 69}
183
+ benches:
184
+ APEX-Agents: 14
185
+ BrowseComp-Plus: 310
186
+ "FrontierSWE v2": 31
187
+ "MCP Atlas": 20
188
+ Multi-SWE-bench: 26
189
+ NL2Repo: 72
190
+ "OSWorld 2": 31
191
+ "SWE-bench Multimodal": 17
192
+ "SWE-bench Pro": 88
193
+ Terminal-Bench: 43
194
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
195
+ pii: null
196
+ pii_evidence: null
197
+ malware: null
198
+ ship_bytes: "yes"
199
+ subset:
200
+ rule: "take all rows; licence, size and version are labels, not filters"
201
+ rows_in_subset: 337638097
202
+ last_built: null
203
+ access:
204
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
205
+ verify: "Request the byte range, which is one gzip-compressed WARC (or ARC) record; strip the WARC and HTTP headers and undo chunked transfer encoding, then compare the sha1 of the HTTP body with hashes.sha1 (hash_scope http_payload). The stored file is the HTTP body."
206
+ hosts: {"s3://eotarchive": 337638097}
207
+ uri_example: "s3://eotarchive/crawl-data/EOT-2004/segments/NARA-000/warc/NARA-PEOT-2004-20041014205819-00000-crawling009-c_NARA-PEOT-2004-20041014205819-00000-crawling009.archive.org.arc.gz"
208
+ needs_auth: {"false": 337638097}
209
+ credential: null
210
+ hash_source: {source_published: 337638097}
211
+ range_supported: null
212
+ reachable:
213
+ index_flag: {unknown: 337638097}
214
+ index_check:
215
+ checked_at: "2026-10-02"
216
+ sample: 30
217
+ checked: 180
218
+ ok: 180
219
+ by_status: {ok: 180}
220
+ our_sample: null
221
+ drift: null
222
+ sample: null
223
+ evidence:
224
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ class: {fill: unfilled, basis: "ships not judged"}
228
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ form.by_ext:
230
+ fill: auto
231
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
232
+ checked_at: "2026-10-03"
233
+ form.date_years:
234
+ fill: auto
235
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
236
+ checked_at: "2026-10-03"
237
+ form.over_cap:
238
+ fill: auto
239
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
240
+ checked_at: "2026-10-03"
241
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
242
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
243
+ form.size_quantiles:
244
+ fill: auto
245
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
246
+ checked_at: "2026-10-03"
247
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
249
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
250
+ provenance.licence_tiers:
251
+ fill: auto
252
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
253
+ checked_at: "2026-10-03"
254
+ provenance.spdx_top:
255
+ fill: auto
256
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
257
+ checked_at: "2026-10-03"
258
+ risk.contamination:
259
+ fill: auto
260
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
261
+ checked_at: "2026-10-03"
262
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
265
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
266
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
268
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
270
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
272
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
273
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
275
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ snapshot.rev:
279
+ fill: auto
280
+ basis: "last commit listed in the publish receipt of this source"
281
+ checked_at: "2026-10-03"
282
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
287
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
290
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ subset:
293
+ fill: auto
294
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
295
+ checked_at: "2026-10-03"
296
+ unit.pointer_count:
297
+ fill: auto
298
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
299
+ checked_at: "2026-10-03"
300
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
301
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
302
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/enron-spreadsheets-emails.yaml ADDED
@@ -0,0 +1,220 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: enron-spreadsheets-emails
8
+ parent: null
9
+ title: "Enron Spreadsheets and Emails (figshare article 1221767)"
10
+ description: "Files of figshare article 1221767: Enron Spreadsheets and Emails (DOI 10.6084/m9.figshare.1221767.v2). One pointer per file."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: "491527a671c38a8627189a36f540bfe7d85a37df", at: "2026-10-01", rev_candidates: 3}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "CC BY 4.0"
17
+ licence_url: "https://doi.org/10.6084/m9.figshare.1221767.v2"
18
+ licence_statement: "License in the figshare article metadata: CC BY 4.0."
19
+ per_collection: null
20
+ training_ok: "yes"
21
+ redistribution_ok: "yes"
22
+ attribution: null
23
+ per_file: null
24
+ licence_tiers: {yes/yes: 14}
25
+ spdx_top: {"(none)": 14}
26
+ spdx_distinct: 1
27
+ licence_sample_check: null
28
+ unit:
29
+ pointer_level: null
30
+ pointer_count: 14
31
+ units: {file: 14}
32
+ state: null
33
+ expanded_count: null
34
+ bytes_total: 9852310070
35
+ rows_without_size: 0
36
+ collections: null
37
+ sub_collections: null
38
+ batches: null
39
+ n_ext: 1
40
+ usable_unit: null
41
+ form:
42
+ by_ext: {"7z": 14}
43
+ bytes_p50: 734003200
44
+ bytes_p95: 992635888
45
+ bytes_max: 992635888
46
+ size_quantiles:
47
+ max: 992635888
48
+ min: 51635782
49
+ p25: 734003200
50
+ p50: 734003200
51
+ p75: 734003200
52
+ p90: 734003200
53
+ p95: 992635888
54
+ p99: 992635888
55
+ p99.9: 992635888
56
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
57
+ by_ext_size:
58
+ - {ext: "7z", rows: 14, bytes: 9852310070, p50: 734003200, p95: 992635888, max: 992635888}
59
+ over_cap:
60
+ cap: 61944279
61
+ rows_over: 13
62
+ bytes_over: 9800674288
63
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
64
+ largest:
65
+ - {pointer_id: "enron-spreadsheets-emails:figshare-1221767:spreadsheets.7z", size_bytes: 992635888}
66
+ - {pointer_id: "enron-spreadsheets-emails:figshare-1221767:eml.7z.012", size_bytes: 734003200}
67
+ - {pointer_id: "enron-spreadsheets-emails:figshare-1221767:eml.7z.011", size_bytes: 734003200}
68
+ - {pointer_id: "enron-spreadsheets-emails:figshare-1221767:eml.7z.010", size_bytes: 734003200}
69
+ - {pointer_id: "enron-spreadsheets-emails:figshare-1221767:eml.7z.009", size_bytes: 734003200}
70
+ date_years: null
71
+ media_type: {"(none)": 14}
72
+ facets: null
73
+ artifact: null
74
+ source_origin: null
75
+ parse_cost: null
76
+ parse_quirks: null
77
+ structure:
78
+ self_describing: null
79
+ integrity_builtin: null
80
+ has_reference_output: null
81
+ invariants: null
82
+ tools: null
83
+ ships:
84
+ problem_statement: {present: null}
85
+ reference_solution: {share: null}
86
+ tests: {present: null}
87
+ env_definition: {present: null}
88
+ known_answer: {present: null}
89
+ generator: {present: null}
90
+ upstream_labels: {present: null}
91
+ class: null
92
+ subject:
93
+ domain_span: null
94
+ shape_span: null
95
+ skill_family_span: null
96
+ ont_domain: null
97
+ tech_vocabulary: null
98
+ risk:
99
+ contamination: none
100
+ overlap_check:
101
+ done: true
102
+ pointers_flagged: 0
103
+ pointers_removed: 0
104
+ by_class: {}
105
+ benches: {}
106
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
107
+ pii: null
108
+ pii_evidence: null
109
+ malware: null
110
+ ship_bytes: "yes"
111
+ subset:
112
+ rule: "take all rows; licence, size and version are labels, not filters"
113
+ rows_in_subset: 14
114
+ last_built:
115
+ date: "2026-10-02"
116
+ rule: "older rule that filtered by licence tier"
117
+ rows: 14
118
+ rows_in_subset: 14
119
+ excluded_by_licence: 0
120
+ duplicates: 0
121
+ access:
122
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
123
+ verify: "Compare size_bytes and md5 (from the figshare API)."
124
+ hosts: {"https://ndownloader.figshare.com": 14}
125
+ uri_example: "https://ndownloader.figshare.com/files/3242534"
126
+ needs_auth: {"false": 14}
127
+ credential: null
128
+ hash_source: {source_published: 14}
129
+ range_supported: null
130
+ reachable:
131
+ index_flag: {"no": 14}
132
+ index_check:
133
+ checked_at: "2026-10-01"
134
+ sample: null
135
+ checked: 14
136
+ ok: 14
137
+ by_status: {ok: 14}
138
+ our_sample: null
139
+ drift: null
140
+ sample: null
141
+ evidence:
142
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
143
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
144
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
145
+ class: {fill: unfilled, basis: "ships not judged"}
146
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
147
+ form.by_ext:
148
+ fill: auto
149
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
150
+ checked_at: "2026-10-03"
151
+ form.date_years:
152
+ fill: auto
153
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
154
+ checked_at: "2026-10-03"
155
+ form.over_cap:
156
+ fill: auto
157
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
158
+ checked_at: "2026-10-03"
159
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
160
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
161
+ form.size_quantiles:
162
+ fill: auto
163
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
164
+ checked_at: "2026-10-03"
165
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
166
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
167
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
168
+ provenance.licence_tiers:
169
+ fill: auto
170
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
171
+ checked_at: "2026-10-03"
172
+ provenance.spdx_top:
173
+ fill: auto
174
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
175
+ checked_at: "2026-10-03"
176
+ risk.contamination:
177
+ fill: auto
178
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
179
+ checked_at: "2026-10-03"
180
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
181
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
184
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
186
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
188
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
189
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
190
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
191
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
192
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
193
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
194
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
195
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
196
+ snapshot.rev:
197
+ fill: auto
198
+ basis: "last commit listed in the publish receipt of this source"
199
+ checked_at: "2026-10-03"
200
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ subset:
211
+ fill: auto
212
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
213
+ checked_at: "2026-10-03"
214
+ unit.pointer_count:
215
+ fill: auto
216
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
217
+ checked_at: "2026-10-03"
218
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
220
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/eu-cellar-documents.yaml ADDED
@@ -0,0 +1,315 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: eu-cellar-documents
8
+ parent: null
9
+ title: "EU Publications Office CELLAR documents"
10
+ description: "Files held in CELLAR, the common data repository of the Publications Office of the European Union (the store behind EUR-Lex and EU Publications): legislation, case law, the Official Journal, publications, meeting minutes and more, in many languages and formats (PDF, Formex XML, XHTML, HTML, DOCX and others). Items are listed through the CELLAR SPARQL endpoint (publications.europa.eu/webapi/rdf/sparql) by the document date of their work. One pointer per item file; uri is the item address, which serves the file on a plain GET. Work id, CELEX number, resource type, language, format, dates, author body and title are in source_meta."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "EU reuse policy (Commission Decision 2011/833/EU), CC BY 4.0"
20
+ licence_url: "https://op.europa.eu/en/web/about-us/legal-notices/publications-office-of-the-european-union-copyright"
21
+ licence_statement: "Publications Office copyright notice: \"Most of the publications on this website are freely available, many of them under Creative Commons licences, and can be downloaded and reproduced provided the source is acknowledged. If you would like to reuse them, please check their respective copyright notices.\" European Commission legal notice: \"Unless otherwise indicated (e.g. in individual copyright notices), content owned by the EU on this website is licensed under the CC BY 4.0 licence\"; reuse is allowed provided appropriate credit is given and changes are indicated. Third-party material needs permission from its owners; patents, trademarks, logos and names are excluded; European Central Bank publications follow their own rules."
22
+ per_collection: null
23
+ training_ok: "yes"
24
+ redistribution_ok: "yes"
25
+ attribution: "Credit the source (© European Union) and indicate any changes."
26
+ per_file: null
27
+ licence_tiers: {yes/yes: 1921815}
28
+ spdx_top: {"(none)": 1921815}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 1921815
34
+ units: {file: 1921815}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 802982306294
38
+ rows_without_size: 1396072
39
+ collections:
40
+ distinct: 22
41
+ top:
42
+ AID_STATE: 96
43
+ COMMUNIC: 175674
44
+ CORRIGENDUM: 35444
45
+ DEC: 208296
46
+ FACTSHEET: 108
47
+ JUDG: 254421
48
+ MINUTES: 76328
49
+ NOTE: 307
50
+ PUB_GEN: 477342
51
+ REG: 693616
52
+ sub_collections: null
53
+ batches: null
54
+ n_ext: 17
55
+ usable_unit: null
56
+ form:
57
+ by_ext:
58
+ pdf: 905771
59
+ xml: 347927
60
+ html: 326852
61
+ jpg: 219659
62
+ xhtml: 60010
63
+ zip: 23641
64
+ tif: 13875
65
+ jpeg: 9410
66
+ doc: 6542
67
+ docx: 5630
68
+ epub: 2206
69
+ png: 245
70
+ xlsx: 22
71
+ txt: 15
72
+ xls: 6
73
+ mobi: 3
74
+ "(none)": 1
75
+ bytes_p50: 59660
76
+ bytes_p95: 2233316
77
+ bytes_max: 2105936403
78
+ size_quantiles:
79
+ max: 2105936403
80
+ min: 97
81
+ p25: 13610
82
+ p50: 59660
83
+ p75: 319903
84
+ p90: 988398
85
+ p95: 2233316
86
+ p99: 23719005
87
+ p99.9: 228772130
88
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
89
+ by_ext_size:
90
+ - {ext: pdf, rows: 905771, bytes: 765046799281, p50: 418668, p95: 9135420, max: 2105936403}
91
+ - {ext: xml, rows: 347927, bytes: 1846490123, p50: 7248, p95: 89228, max: 5531620}
92
+ - {ext: html, rows: 326852, bytes: 22207167150, p50: 19484, p95: 295521, max: 181143123}
93
+ - {ext: jpg, rows: 219659, bytes: 1076232823, p50: 8886, p95: 409866, max: 2365783}
94
+ - {ext: xhtml, rows: 60010, bytes: 782026915, p50: 76222, p95: 244306, max: 2391882}
95
+ - {ext: zip, rows: 23641, bytes: 4536212421, p50: 10102, p95: 127061, max: 535443097}
96
+ - {ext: tif, rows: 13875, bytes: 3545274298, p50: 53665, p95: 5739068, max: 44609816}
97
+ - {ext: jpeg, rows: 9410, bytes: 226638919, p50: 23802, p95: 37055, max: 72310}
98
+ - {ext: doc, rows: 6542, bytes: 827691024, p50: 133120, p95: 601088, max: 4636160}
99
+ - {ext: docx, rows: 5630, bytes: 1130250213, p50: 98554, p95: 989263, max: 9228165}
100
+ - {ext: epub, rows: 2206, bytes: 1752043114, p50: 3880968, p95: 24663215, max: 279129453}
101
+ - {ext: png, rows: 245, bytes: 231336, p50: 3866, p95: 209025, max: 209025}
102
+ - {ext: xlsx, rows: 22, bytes: 2192571, p50: 35371, p95: 349034, max: 418345}
103
+ - {ext: txt, rows: 15, bytes: 0, p50: null, p95: null, max: null}
104
+ - {ext: xls, rows: 6, bytes: 820736, p50: 82944, p95: 336384, max: 336384}
105
+ - {ext: mobi, rows: 3, bytes: 2235370, p50: 756983, p95: 770813, max: 770813}
106
+ - {ext: "(none)", rows: 1, bytes: 0, p50: null, p95: null, max: null}
107
+ over_cap:
108
+ cap: 61944279
109
+ rows_over: 2017
110
+ bytes_over: 414408649123
111
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
112
+ largest:
113
+ - pointer_id: "eu-cellar-documents:PUB_GEN:eba62f06-de62-4430-a4f8-0671371bfe5c.0002.03/DOC_1"
114
+ size_bytes: 2105936403
115
+ - pointer_id: "eu-cellar-documents:PUB_GEN:51f44859-1ed5-4c25-ba6f-86c43d38299f.0001.02/DOC_2"
116
+ size_bytes: 2102069361
117
+ - pointer_id: "eu-cellar-documents:PUB_GEN:eba62f06-de62-4430-a4f8-0671371bfe5c.0001.03/DOC_1"
118
+ size_bytes: 2100788122
119
+ - pointer_id: "eu-cellar-documents:PUB_GEN:eba62f06-de62-4430-a4f8-0671371bfe5c.0003.04/DOC_1"
120
+ size_bytes: 2093731595
121
+ - pointer_id: "eu-cellar-documents:PUB_GEN:bfe027a3-427c-4288-bf11-08af8d282388.0001.02/DOC_2"
122
+ size_bytes: 2014669688
123
+ date_years: null
124
+ media_type:
125
+ application/epub+zip: 2206
126
+ application/msword: 6533
127
+ application/pdf: 905679
128
+ application/vnd.openxmlformats-officedocument.wordprocessingml.document.main+xml: 5624
129
+ application/xhtml+xml: 154508
130
+ application/xml: 355530
131
+ application/zip: 23647
132
+ image/jpeg: 229064
133
+ image/tiff: 13843
134
+ text/html: 224748
135
+ facets:
136
+ format:
137
+ doc: 6542
138
+ docx: 5658
139
+ epub: 3762
140
+ fmx4: 373073
141
+ html: 191468
142
+ leaflet: 4031
143
+ pdf: 391112
144
+ pdf1x: 645
145
+ pdfa1a: 74606
146
+ pdfa1b: 388967
147
+ pdfa2a: 34880
148
+ pdfx: 86323
149
+ pdfx4: 7765
150
+ poster_prom: 1883
151
+ print: 65194
152
+ print_losl: 1242
153
+ print_sftcv: 20910
154
+ print_stpl: 24636
155
+ xhtml: 216629
156
+ xml: 19969
157
+ is_meeting_record: {"false": 1845487, "true": 76328}
158
+ mime:
159
+ application/epub+zip: 2206
160
+ application/msword: 6533
161
+ application/pdf: 418607
162
+ "application/pdf;type=pdf1x": 1070
163
+ "application/pdf;type=pdfa1a": 65066
164
+ "application/pdf;type=pdfa1b": 386445
165
+ "application/pdf;type=pdfa2a": 10661
166
+ "application/pdf;type=pdfa2b": 69
167
+ "application/pdf;type=pdfx": 23761
168
+ application/xhtml+xml: 154508
169
+ application/xml: 19946
170
+ "application/xml;type=fmx4": 335584
171
+ application/zip: 23509
172
+ "application/zip;type=akn4eu": 138
173
+ image/jpeg: 229064
174
+ image/png: 240
175
+ image/tiff: 13843
176
+ text/html: 189733
177
+ "text/html;type=simplified": 35015
178
+ text/plain: 23
179
+ artifact: null
180
+ source_origin: null
181
+ parse_cost: null
182
+ parse_quirks: null
183
+ structure:
184
+ self_describing: null
185
+ integrity_builtin: null
186
+ has_reference_output: null
187
+ invariants: null
188
+ tools: null
189
+ ships:
190
+ problem_statement: {present: null}
191
+ reference_solution: {share: null}
192
+ tests: {present: null}
193
+ env_definition: {present: null}
194
+ known_answer: {present: null}
195
+ generator: {present: null}
196
+ upstream_labels: {present: null}
197
+ class: null
198
+ subject:
199
+ domain_span: null
200
+ shape_span: null
201
+ skill_family_span: null
202
+ ont_domain: null
203
+ tech_vocabulary: null
204
+ risk:
205
+ contamination: none
206
+ overlap_check:
207
+ done: true
208
+ pointers_flagged: 0
209
+ pointers_removed: 0
210
+ by_class: {}
211
+ benches: {}
212
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
213
+ pii: null
214
+ pii_evidence: null
215
+ malware: null
216
+ ship_bytes: "yes"
217
+ subset:
218
+ rule: "take all rows; licence, size and version are labels, not filters"
219
+ rows_in_subset: 1921815
220
+ last_built: null
221
+ access:
222
+ method: "GET uri (the https item address; alt_uris holds the http form). No login or content negotiation is needed."
223
+ verify: "Compare size_bytes and sha1 when given (sha1 is the SHA-1 digest CELLAR publishes for the item). Rows with a null size or without hashes (computed_on_fetch) count as ok when the download succeeds; record the measured size and sha256."
224
+ hosts: {"https://publications.europa.eu": 1921815}
225
+ uri_example: "https://publications.europa.eu/resource/cellar/1b959c82-e69e-449d-bf44-20df4b02c51f.0003.01/DOC_1"
226
+ needs_auth: {"false": 1921815}
227
+ credential: null
228
+ hash_source: {computed_on_fetch: 1396295, source_published: 525520}
229
+ range_supported: null
230
+ reachable:
231
+ index_flag: {unknown: 1921815}
232
+ index_check: null
233
+ our_sample: null
234
+ drift: null
235
+ sample: null
236
+ evidence:
237
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
238
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
239
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
240
+ class: {fill: unfilled, basis: "ships not judged"}
241
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
242
+ form.by_ext:
243
+ fill: auto
244
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
245
+ checked_at: "2026-10-03"
246
+ form.date_years:
247
+ fill: auto
248
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
249
+ checked_at: "2026-10-03"
250
+ form.over_cap:
251
+ fill: auto
252
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
253
+ checked_at: "2026-10-03"
254
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ form.size_quantiles:
257
+ fill: auto
258
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
259
+ checked_at: "2026-10-03"
260
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
261
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
262
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ provenance.licence_tiers:
264
+ fill: auto
265
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
266
+ checked_at: "2026-10-03"
267
+ provenance.spdx_top:
268
+ fill: auto
269
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
270
+ checked_at: "2026-10-03"
271
+ risk.contamination:
272
+ fill: auto
273
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
274
+ checked_at: "2026-10-03"
275
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
279
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
287
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
290
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ snapshot.rev:
292
+ fill: auto
293
+ basis: "last commit listed in the publish receipt of this source"
294
+ checked_at: "2026-10-03"
295
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
296
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
297
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
298
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
299
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
300
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
301
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
302
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
303
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
304
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
305
+ subset:
306
+ fill: auto
307
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
308
+ checked_at: "2026-10-03"
309
+ unit.pointer_count:
310
+ fill: auto
311
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
312
+ checked_at: "2026-10-03"
313
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
314
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
315
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/european-parliament-plenary.yaml ADDED
@@ -0,0 +1,270 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: european-parliament-plenary
8
+ parent: null
9
+ title: "European Parliament plenary records"
10
+ description: "Plenary records from the European Parliament Open Data API v2 (data.europarl.europa.eu): plenary speeches from July 2019 on (one pointer per speech per language; the hash covers the speech's XML fragment), plenary documents, and plenary session documents such as the verbatim report of proceedings (one metadata pointer per document plus one pointer per file, such as XML, PDF or DOCX). Sitting date, speaker, language and document type are in source_meta."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-02"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Creative Commons Attribution 4.0 International (CC BY 4.0)"
20
+ licence_url: "https://eur-lex.europa.eu/eli/C/2025/341/oj/eng"
21
+ licence_statement: "Article 4 of Bureau Decision C/2025/341: the re-use of Parliament's open data is subject to the conditions of CC BY 4.0; raw data, metadata or comparable documents may alternatively be distributed under CC0 1.0. The Parliament's legal notice (https://www.europarl.europa.eu/legal-notice/en/) allows reuse for personal, non-commercial or commercial purposes, provided the entire item is reproduced and the source is acknowledged; interpretation of speeches is not an official or verbatim text."
22
+ per_collection: null
23
+ training_ok: "yes"
24
+ redistribution_ok: "yes"
25
+ attribution: "Acknowledge the source: © European Union, [year(s)], Source: European Parliament"
26
+ per_file: null
27
+ licence_tiers: {yes/yes: 1233677}
28
+ spdx_top: {"(none)": 1233677}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 1233677
34
+ units: {"API response": 104530, file: 1129147}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 38009011404
38
+ rows_without_size: 712457
39
+ collections:
40
+ distinct: 3
41
+ top: {plenary-documents: 1146206, plenary-session-documents: 1039, speeches: 86432}
42
+ sub_collections: null
43
+ batches: null
44
+ n_ext: 7
45
+ usable_unit: null
46
+ form:
47
+ by_ext: {pdf: 346444, xml: 312003, doc: 175436, docx: 171545, html: 163876, xhtml: 46275, json: 18098}
48
+ bytes_p50: 48081
49
+ bytes_p95: 247741
50
+ bytes_max: 6737815
51
+ size_quantiles:
52
+ max: 6737815
53
+ min: 0
54
+ p25: 9581
55
+ p50: 48081
56
+ p75: 72431
57
+ p90: 161116
58
+ p95: 247741
59
+ p99: 471177
60
+ p99.9: 1747169
61
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
62
+ by_ext_size:
63
+ - {ext: pdf, rows: 346444, bytes: 23291462186, p50: 127706, p95: 401587, max: 6737815}
64
+ - {ext: xml, rows: 312003, bytes: 1326898089, p50: 2161, p95: 25850, max: 6130209}
65
+ - {ext: doc, rows: 175436, bytes: 2516283150, p50: 60416, p95: 117248, max: 2544128}
66
+ - {ext: docx, rows: 171545, bytes: 7704792766, p50: 52086, p95: 160149, max: 1271852}
67
+ - {ext: html, rows: 163876, bytes: 1425845335, p50: 22235, p95: 49089, max: 3394456}
68
+ - {ext: xhtml, rows: 46275, bytes: 1076228912, p50: 11101, p95: 76059, max: 2392544}
69
+ - {ext: json, rows: 18098, bytes: 667500966, p50: 40627, p95: 58710, max: 282548}
70
+ over_cap:
71
+ cap: 61944279
72
+ rows_over: 0
73
+ bytes_over: 0
74
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
75
+ largest:
76
+ - pointer_id: "european-parliament-plenary:plenary-session-documents:CRE-9-2023-05-10/CRE-9-2023-05-10_mul.pdf"
77
+ size_bytes: 6737815
78
+ - pointer_id: "european-parliament-plenary:plenary-session-documents:CRE-9-2024-04-24/CRE-9-2024-04-24_mul.pdf"
79
+ size_bytes: 6452423
80
+ - pointer_id: "european-parliament-plenary:plenary-documents:A-8-2019-0200/A-8-2019-0200_el.xml"
81
+ size_bytes: 6130209
82
+ - pointer_id: "european-parliament-plenary:plenary-documents:A-8-2019-0200/A-8-2019-0200_bg.xml"
83
+ size_bytes: 6075140
84
+ - pointer_id: "european-parliament-plenary:plenary-session-documents:CRE-9-2022-05-04/CRE-9-2022-05-04_mul.pdf"
85
+ size_bytes: 6029111
86
+ date_years: null
87
+ media_type:
88
+ application/json: 18098
89
+ application/msword: 175393
90
+ application/pdf: 334917
91
+ application/vnd.openxmlformats-officedocument.wordprocessingml.document: 160061
92
+ application/xhtml+xml: 46275
93
+ application/xml: 311852
94
+ pdf: 11527
95
+ text/html: 163876
96
+ vnd.openxmlformats-officedocument.wordprocessingml.document: 11484
97
+ xml: 151
98
+ facets:
99
+ activity_type:
100
+ MEETING_NAME_CHAIR: 1499
101
+ PLENARY_DEBATE_SPEECH: 76018
102
+ PLENARY_DEBATE_WRITTEN_STATEMENT: 3214
103
+ PROCEEDING_ACTIVITY: 5701
104
+ document_type:
105
+ CRE_PLENARY: 1039
106
+ QUESTION_RESOLUTION_MOTION: 33135
107
+ REPORT_PLENARY: 560768
108
+ RESOLUTION_MOTION: 505900
109
+ RESOLUTION_MOTION_JOINT: 46403
110
+ format: {DOC: 175436, DOCX: 171545, HTML: 163876, PDF: 346444, XHTML: 46275, XML: 225571}
111
+ kind: {file: 1129147, metadata: 18098, speech: 86432}
112
+ language:
113
+ cs: 50521
114
+ da: 50152
115
+ de: 59884
116
+ el: 51369
117
+ en: 98167
118
+ es: 55965
119
+ et: 49299
120
+ fi: 50533
121
+ fr: 58963
122
+ hu: 50614
123
+ it: 56101
124
+ lt: 49817
125
+ lv: 50054
126
+ nl: 51376
127
+ pl: 53658
128
+ pt: 52478
129
+ ro: 49800
130
+ sk: 50779
131
+ sl: 49701
132
+ sv: 50681
133
+ parliamentary_term: {ep-10: 62401, ep-5: 146, ep-6: 137461, ep-7: 213699, ep-8: 544803, ep-9: 188703}
134
+ artifact: null
135
+ source_origin: null
136
+ parse_cost: null
137
+ parse_quirks: null
138
+ structure:
139
+ self_describing: null
140
+ integrity_builtin: null
141
+ has_reference_output: null
142
+ invariants: null
143
+ tools: null
144
+ ships:
145
+ problem_statement: {present: null}
146
+ reference_solution: {share: null}
147
+ tests: {present: null}
148
+ env_definition: {present: null}
149
+ known_answer: {present: null}
150
+ generator: {present: null}
151
+ upstream_labels: {present: null}
152
+ class: null
153
+ subject:
154
+ domain_span: null
155
+ shape_span: null
156
+ skill_family_span: null
157
+ ont_domain: null
158
+ tech_vocabulary: null
159
+ risk:
160
+ contamination: none
161
+ overlap_check:
162
+ done: true
163
+ pointers_flagged: 0
164
+ pointers_removed: 0
165
+ by_class: {}
166
+ benches: {}
167
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
168
+ pii: null
169
+ pii_evidence: null
170
+ malware: null
171
+ ship_bytes: "yes"
172
+ subset:
173
+ rule: "take all rows; licence, size and version are labels, not filters"
174
+ rows_in_subset: 1233677
175
+ last_built: null
176
+ access:
177
+ method: "GET uri. API rows (ext json or xml) return JSON-LD (format=application/ld+json is in the uri); file rows return the file itself."
178
+ verify: "API rows: take the field named in source_meta.api_body_field (a string as is, anything else serialised with json.dumps(ensure_ascii=False)), encode it as UTF-8 and compare size and sha256; the stored file is these bytes. File rows (computed_on_fetch): compare size with size_bytes (the source's byteSize) and record the measured sha256."
179
+ hosts: {"https://data.europarl.europa.eu": 1233677}
180
+ uri_example: "https://data.europarl.europa.eu/api/v2/plenary-documents/A-10-2024-0001?format=application%2Fld%2Bjson"
181
+ needs_auth: {"false": 1233677}
182
+ credential: null
183
+ hash_source: {computed: 104530, computed_on_fetch: 1129147}
184
+ range_supported: null
185
+ reachable:
186
+ index_flag: {unknown: 1233677}
187
+ index_check: null
188
+ our_sample: null
189
+ drift: null
190
+ sample: null
191
+ evidence:
192
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
193
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
194
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
195
+ class: {fill: unfilled, basis: "ships not judged"}
196
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
197
+ form.by_ext:
198
+ fill: auto
199
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
200
+ checked_at: "2026-10-03"
201
+ form.date_years:
202
+ fill: auto
203
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
204
+ checked_at: "2026-10-03"
205
+ form.over_cap:
206
+ fill: auto
207
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
208
+ checked_at: "2026-10-03"
209
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ form.size_quantiles:
212
+ fill: auto
213
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
214
+ checked_at: "2026-10-03"
215
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
217
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ provenance.licence_tiers:
219
+ fill: auto
220
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
221
+ checked_at: "2026-10-03"
222
+ provenance.spdx_top:
223
+ fill: auto
224
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
225
+ checked_at: "2026-10-03"
226
+ risk.contamination:
227
+ fill: auto
228
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
229
+ checked_at: "2026-10-03"
230
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
234
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
237
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
238
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
239
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
240
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
241
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
242
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
243
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
244
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
245
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ snapshot.rev:
247
+ fill: auto
248
+ basis: "last commit listed in the publish receipt of this source"
249
+ checked_at: "2026-10-03"
250
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
251
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
252
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
253
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
257
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
260
+ subset:
261
+ fill: auto
262
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
263
+ checked_at: "2026-10-03"
264
+ unit.pointer_count:
265
+ fill: auto
266
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
267
+ checked_at: "2026-10-03"
268
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
270
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/f-droid-apps.yaml ADDED
@@ -0,0 +1,302 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: f-droid-apps
8
+ parent: null
9
+ title: "F-Droid apps"
10
+ description: "Android apps from F-Droid (f-droid.org), a repository of free and open source Android apps: for every version of every app, one pointer to the APK and one to the source tarball F-Droid built it from (when the index lists one). Covers the main repository and the archive repository (older versions). Versions are read from the repository index (entry.json and index-v2.json); sizes and SHA-256 come from the index, licenses from each app's metadata (repo index sha256 97ae2959629f8ad3f8dfc62a1c8667397cdc89ecdab59f894e236783efd1dc73; archive index sha256 ed1c472275fc89e1082db85b9dc62a440ca5e90a64e459acef0f1fb5a91967a6). Packages can contain malicious code; treat files as data and run nothing outside a sandbox."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per app (license field of the app metadata)"
20
+ licence_url: "https://f-droid.org/docs/Inclusion_Policy/"
21
+ licence_statement: "F-Droid gives no license of its own; each app declares its license (SPDX). Its inclusion policy says all apps in the repository must be free, libre and open source software."
22
+ per_collection: null
23
+ training_ok: conditional
24
+ redistribution_ok: conditional
25
+ attribution: null
26
+ per_file: "license_override.notes gives the license field of the app; license_override.spdx gives the SPDX expression where it is recognised. The APK and the source tarball of a version share the app's license. Values the license table does not recognise (such as PublicDomain, WTFPL, BSD-4-Clause or X11) and a missing field are marked unknown."
27
+ licence_tiers: {unknown/unknown: 232, yes/conditional: 140279, yes/yes: 55865}
28
+ spdx_top:
29
+ "(none)": 232
30
+ "0BSD": 74
31
+ AAL: 134
32
+ AGPL-3.0-only: 10738
33
+ AGPL-3.0-or-later: 5498
34
+ Apache-2.0: 26631
35
+ Artistic-2.0: 44
36
+ BSD-2-Clause: 568
37
+ BSD-3-Clause: 2993
38
+ CC-BY-4.0: 98
39
+ CC-BY-SA-4.0: 66
40
+ CC0-1.0: 184
41
+ CDDL-1.0: 58
42
+ CECILL-2.1: 18
43
+ EPL-1.0: 18
44
+ spdx_distinct: 38
45
+ licence_sample_check: null
46
+ unit:
47
+ pointer_level: null
48
+ pointer_count: 196376
49
+ units: {file: 196376}
50
+ state: null
51
+ expanded_count: null
52
+ bytes_total: 7297488209594
53
+ rows_without_size: 0
54
+ collections:
55
+ distinct: 6769
56
+ top:
57
+ com.adilhanney.saber: 640
58
+ com.ismartcoding.plain: 894
59
+ com.nextcloud.android.beta: 2436
60
+ com.presley.flexify: 704
61
+ com.tutpro.baresip: 652
62
+ com.unciv.app: 1334
63
+ eu.faircode.email: 1544
64
+ eu.siacs.conversations: 978
65
+ org.mozilla.fennec_fdroid: 981
66
+ org.openhab.habdroid.beta: 742
67
+ sub_collections: null
68
+ batches: null
69
+ n_ext: 3
70
+ usable_unit: null
71
+ form:
72
+ by_ext: {apk: 98291, gz: 98082, zip: 3}
73
+ bytes_p50: 7007074
74
+ bytes_p95: 167087182
75
+ bytes_max: 2479783254
76
+ size_quantiles:
77
+ max: 2479783254
78
+ min: 258
79
+ p25: 2135966
80
+ p50: 7007074
81
+ p75: 26525010
82
+ p90: 80655830
83
+ p95: 167087182
84
+ p99: 495099852
85
+ p99.9: 979172551
86
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
87
+ by_ext_size:
88
+ - {ext: apk, rows: 98291, bytes: 1986875644469, p50: 8312500, p95: 78281150, max: 1494811148}
89
+ - {ext: gz, rows: 98082, bytes: 5310594626978, p50: 5519058, p95: 370665536, max: 2479783254}
90
+ - {ext: zip, rows: 3, bytes: 17938147, p50: 5888840, p95: 6303750, max: 6303750}
91
+ over_cap:
92
+ cap: 61944279
93
+ rows_over: 25349
94
+ bytes_over: 5394445948158
95
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
96
+ largest:
97
+ - pointer_id: "f-droid-apps:org.commonvoice.saverio:repo/org.commonvoice.saverio_182_src.tar.gz"
98
+ size_bytes: 2479783254
99
+ - pointer_id: "f-droid-apps:org.commonvoice.saverio:repo/org.commonvoice.saverio_183_src.tar.gz"
100
+ size_bytes: 2479740190
101
+ - pointer_id: "f-droid-apps:org.commonvoice.saverio:repo/org.commonvoice.saverio_179_src.tar.gz"
102
+ size_bytes: 2479419545
103
+ - pointer_id: "f-droid-apps:org.commonvoice.saverio:archive/org.commonvoice.saverio_174_src.tar.gz"
104
+ size_bytes: 2479161338
105
+ - pointer_id: "f-droid-apps:org.commonvoice.saverio:archive/org.commonvoice.saverio_175_src.tar.gz"
106
+ size_bytes: 2479155291
107
+ date_years: null
108
+ media_type: {"(none)": 196376}
109
+ facets:
110
+ anti_features:
111
+ Ads: 366
112
+ "Ads; NonFreeNet": 728
113
+ "DisabledAlgorithm; KnownVuln": 234
114
+ KnownVuln: 404
115
+ "KnownVuln; Tracking": 599
116
+ NoSourceSince: 3722
117
+ "NoSourceSince; NonFreeNet": 582
118
+ NonFreeAdd: 703
119
+ "NonFreeAdd; NonFreeNet": 220
120
+ NonFreeAssets: 1326
121
+ "NonFreeAssets; NonFreeNet": 312
122
+ "NonFreeAssets; NonFreeNet; Tracking": 218
123
+ NonFreeDep: 412
124
+ NonFreeNet: 26559
125
+ "NonFreeNet; TetheredNet": 1220
126
+ "NonFreeNet; TetheredNet; Tracking": 974
127
+ "NonFreeNet; Tracking": 2966
128
+ TetheredNet: 3302
129
+ "TetheredNet; Tracking": 584
130
+ Tracking: 2440
131
+ fdroid_repo: {archive: 168784, repo: 27592}
132
+ file_kind: {apk: 98294, src: 98082}
133
+ license:
134
+ AAL: 134
135
+ AGPL-3.0-only: 10738
136
+ AGPL-3.0-or-later: 5498
137
+ Apache-2.0: 26631
138
+ BSD-2-Clause: 568
139
+ BSD-3-Clause: 2993
140
+ CC0-1.0: 184
141
+ EPL-2.0: 872
142
+ EUPL-1.2: 486
143
+ GPL-2.0-only: 5729
144
+ GPL-2.0-or-later: 3878
145
+ GPL-3.0-only: 71455
146
+ GPL-3.0-or-later: 34452
147
+ ISC: 454
148
+ LGPL-2.1-only: 592
149
+ LGPL-3.0-only: 606
150
+ LGPL-3.0-or-later: 160
151
+ MIT: 23907
152
+ MPL-2.0: 5505
153
+ Unlicense: 768
154
+ native_code:
155
+ arm64-v8a: 14288
156
+ "arm64-v8a; armeabi-v7a": 3282
157
+ "arm64-v8a; armeabi-v7a; x86": 430
158
+ "arm64-v8a; armeabi-v7a; x86; x86_64": 48287
159
+ "arm64-v8a; armeabi-v7a; x86_64": 6324
160
+ "arm64-v8a; armeabi; armeabi-v7a; x86": 162
161
+ "arm64-v8a; armeabi; x86": 174
162
+ "arm64-v8a; armeabi; x86; x86_64": 326
163
+ "arm64-v8a; x86_64": 400
164
+ armeabi: 658
165
+ armeabi-v7a: 12013
166
+ "armeabi-v7a; x86": 576
167
+ "armeabi; armeabi-v7a": 200
168
+ "armeabi; armeabi-v7a; mips; x86": 745
169
+ "armeabi; armeabi-v7a; x86": 654
170
+ "armeabi; mips; x86": 200
171
+ "armeabi; x86": 216
172
+ x86: 4503
173
+ "x86; x86_64": 122
174
+ x86_64: 9288
175
+ artifact: null
176
+ source_origin: null
177
+ parse_cost: null
178
+ parse_quirks: null
179
+ structure:
180
+ self_describing: null
181
+ integrity_builtin: null
182
+ has_reference_output: null
183
+ invariants: null
184
+ tools: null
185
+ ships:
186
+ problem_statement: {present: null}
187
+ reference_solution: {share: null}
188
+ tests: {present: null}
189
+ env_definition: {present: null}
190
+ known_answer: {present: null}
191
+ generator: {present: null}
192
+ upstream_labels: {present: null}
193
+ class: null
194
+ subject:
195
+ domain_span: null
196
+ shape_span: null
197
+ skill_family_span: null
198
+ ont_domain: null
199
+ tech_vocabulary: null
200
+ risk:
201
+ contamination: null
202
+ overlap_check: null
203
+ pii: null
204
+ pii_evidence: null
205
+ malware: null
206
+ ship_bytes: conditional
207
+ subset:
208
+ rule: "take all rows; licence, size and version are labels, not filters"
209
+ rows_in_subset: 196376
210
+ last_built: null
211
+ access:
212
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
213
+ verify: "Compare size_bytes and sha256 (published in index-v2.json)."
214
+ hosts: {"https://f-droid.org": 196376}
215
+ uri_example: "https://f-droid.org/archive/An.stop_10.apk"
216
+ needs_auth: {"false": 196376}
217
+ credential: null
218
+ hash_source: {source_published: 196376}
219
+ range_supported: null
220
+ reachable:
221
+ index_flag: {"no": 196376}
222
+ index_check: null
223
+ our_sample: null
224
+ drift: null
225
+ sample: null
226
+ evidence:
227
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
228
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ class: {fill: unfilled, basis: "ships not judged"}
231
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ form.by_ext:
233
+ fill: auto
234
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
235
+ checked_at: "2026-10-03"
236
+ form.date_years:
237
+ fill: auto
238
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
239
+ checked_at: "2026-10-03"
240
+ form.over_cap:
241
+ fill: auto
242
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
243
+ checked_at: "2026-10-03"
244
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
245
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ form.size_quantiles:
247
+ fill: auto
248
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
249
+ checked_at: "2026-10-03"
250
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
251
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
252
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
253
+ provenance.licence_tiers:
254
+ fill: auto
255
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
256
+ checked_at: "2026-10-03"
257
+ provenance.spdx_top:
258
+ fill: auto
259
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
260
+ checked_at: "2026-10-03"
261
+ risk.contamination: {fill: unfilled, basis: "this source was not part of the benchmark overlap scan"}
262
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
265
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
266
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
268
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
270
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
272
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
273
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
275
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ snapshot.rev:
279
+ fill: auto
280
+ basis: "last commit listed in the publish receipt of this source"
281
+ checked_at: "2026-10-03"
282
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
287
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
290
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ subset:
293
+ fill: auto
294
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
295
+ checked_at: "2026-10-03"
296
+ unit.pointer_count:
297
+ fill: auto
298
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
299
+ checked_at: "2026-10-03"
300
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
301
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
302
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/fatcat-files.yaml ADDED
@@ -0,0 +1,213 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: fatcat-files
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers: {unknown/unknown: 957833}
29
+ spdx_top: {"(none)": 957833}
30
+ spdx_distinct: 1
31
+ licence_sample_check: null
32
+ unit:
33
+ pointer_level: null
34
+ pointer_count: 957833
35
+ units: {file: 957833}
36
+ state: null
37
+ expanded_count: null
38
+ bytes_total: 1639830748652
39
+ rows_without_size: 0
40
+ collections: null
41
+ sub_collections: null
42
+ batches: null
43
+ n_ext: 1
44
+ usable_unit: null
45
+ form:
46
+ by_ext: {pdf: 957833}
47
+ bytes_p50: 641290
48
+ bytes_p95: 6282309
49
+ bytes_max: 1381424460
50
+ size_quantiles:
51
+ max: 1381424460
52
+ min: 1923
53
+ p25: 307853
54
+ p50: 641290
55
+ p75: 1465167
56
+ p90: 3582312
57
+ p95: 6282309
58
+ p99: 18176681
59
+ p99.9: 54219062
60
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
61
+ by_ext_size:
62
+ - {ext: pdf, rows: 957833, bytes: 1639830748652, p50: 641290, p95: 6282309, max: 1381424460}
63
+ over_cap:
64
+ cap: 61944279
65
+ rows_over: 705
66
+ bytes_over: 69566785413
67
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
68
+ largest:
69
+ - {pointer_id: "fatcat-files:pdf:jyqvtpzgcvhslizr7vgrf6llle", size_bytes: 1381424460}
70
+ - {pointer_id: "fatcat-files:pdf:m32l2eugqzafhb4i7rqokbqbfm", size_bytes: 1370635190}
71
+ - {pointer_id: "fatcat-files:pdf:ncaddtbvkffs7hskr5puzxgw5u", size_bytes: 327678970}
72
+ - {pointer_id: "fatcat-files:pdf:dhlemzoy2zfevaspzkptakxjfu", size_bytes: 311978481}
73
+ - {pointer_id: "fatcat-files:pdf:x2jvdcoj3ngw5lu7ihqyjkoyoq", size_bytes: 304247187}
74
+ date_years: null
75
+ media_type: {application/pdf: 957833}
76
+ facets: null
77
+ artifact: null
78
+ source_origin: null
79
+ parse_cost: null
80
+ parse_quirks: null
81
+ structure:
82
+ self_describing: null
83
+ integrity_builtin: null
84
+ has_reference_output: null
85
+ invariants: null
86
+ tools: null
87
+ ships:
88
+ problem_statement: {present: null}
89
+ reference_solution: {share: null}
90
+ tests: {present: null}
91
+ env_definition: {present: null}
92
+ known_answer: {present: null}
93
+ generator: {present: null}
94
+ upstream_labels: {present: null}
95
+ class: null
96
+ subject:
97
+ domain_span: null
98
+ shape_span: null
99
+ skill_family_span: null
100
+ ont_domain: null
101
+ tech_vocabulary: null
102
+ risk:
103
+ contamination: none
104
+ overlap_check:
105
+ done: true
106
+ pointers_flagged: 0
107
+ pointers_removed: 0
108
+ by_class: {}
109
+ benches: {}
110
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
111
+ pii: null
112
+ pii_evidence: null
113
+ malware: null
114
+ ship_bytes: unknown
115
+ subset:
116
+ rule: "take all rows; licence, size and version are labels, not filters"
117
+ rows_in_subset: 957833
118
+ last_built: null
119
+ access:
120
+ method: null
121
+ verify: null
122
+ hosts: {"https://web.archive.org": 957833}
123
+ uri_example: "https://web.archive.org/web/20191027035422id_/https://royalsocietypublishing.org/doi/pdf/10.1098/rstb.2012.0109"
124
+ needs_auth: {"false": 957833}
125
+ credential: null
126
+ hash_source: {source_published: 957833}
127
+ range_supported: null
128
+ reachable:
129
+ index_flag: {"(none)": 957833}
130
+ index_check: null
131
+ our_sample: null
132
+ drift: null
133
+ sample: null
134
+ evidence:
135
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
136
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
137
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
138
+ class: {fill: unfilled, basis: "ships not judged"}
139
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
140
+ form.by_ext:
141
+ fill: auto
142
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
143
+ checked_at: "2026-10-03"
144
+ form.date_years:
145
+ fill: auto
146
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
147
+ checked_at: "2026-10-03"
148
+ form.over_cap:
149
+ fill: auto
150
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
151
+ checked_at: "2026-10-03"
152
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
153
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
154
+ form.size_quantiles:
155
+ fill: auto
156
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
157
+ checked_at: "2026-10-03"
158
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
159
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
160
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
161
+ provenance.licence_tiers:
162
+ fill: auto
163
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
164
+ checked_at: "2026-10-03"
165
+ provenance.spdx_top:
166
+ fill: auto
167
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
168
+ checked_at: "2026-10-03"
169
+ risk.contamination:
170
+ fill: auto
171
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
172
+ checked_at: "2026-10-03"
173
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
174
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
175
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
176
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
177
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
178
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
179
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
180
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
181
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
184
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
186
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
188
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
189
+ snapshot.rev:
190
+ fill: auto
191
+ basis: "last commit listed in the publish receipt of this source"
192
+ checked_at: "2026-10-03"
193
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
194
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
195
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
196
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
197
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
198
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
199
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ subset:
204
+ fill: auto
205
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
206
+ checked_at: "2026-10-03"
207
+ unit.pointer_count:
208
+ fill: auto
209
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
210
+ checked_at: "2026-10-03"
211
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
212
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
213
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/fedora-koji-build-logs.yaml ADDED
@@ -0,0 +1,234 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: fedora-koji-build-logs
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers: {unknown/unknown: 20284}
29
+ spdx_top: {"(none)": 20284}
30
+ spdx_distinct: 1
31
+ licence_sample_check: null
32
+ unit:
33
+ pointer_level: null
34
+ pointer_count: 20284
35
+ units: {file: 20284}
36
+ state: null
37
+ expanded_count: null
38
+ bytes_total: 31376260955
39
+ rows_without_size: 0
40
+ collections:
41
+ distinct: 2
42
+ top: {complete: 17207, failed: 3077}
43
+ sub_collections: null
44
+ batches: null
45
+ n_ext: 1
46
+ usable_unit: null
47
+ form:
48
+ by_ext: {log: 20284}
49
+ bytes_p50: 108899
50
+ bytes_p95: 4534557
51
+ bytes_max: 538146198
52
+ size_quantiles:
53
+ max: 538146198
54
+ min: 18
55
+ p25: 11761
56
+ p50: 108899
57
+ p75: 454075
58
+ p90: 1536072
59
+ p95: 4534557
60
+ p99: 26316507
61
+ p99.9: 145554630
62
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
63
+ by_ext_size:
64
+ - {ext: log, rows: 20284, bytes: 31376260955, p50: 108899, p95: 4534557, max: 538146198}
65
+ over_cap:
66
+ cap: 61944279
67
+ rows_over: 103
68
+ bytes_over: 12526905525
69
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
70
+ largest:
71
+ - {pointer_id: "fedora-koji-build-logs:complete:3111988/ppc64le/build.log", size_bytes: 538146198}
72
+ - {pointer_id: "fedora-koji-build-logs:complete:3111988/aarch64/build.log", size_bytes: 454782717}
73
+ - {pointer_id: "fedora-koji-build-logs:complete:3111988/x86_64/build.log", size_bytes: 321934226}
74
+ - {pointer_id: "fedora-koji-build-logs:complete:3108978/x86_64/build.log", size_bytes: 284654282}
75
+ - {pointer_id: "fedora-koji-build-logs:complete:3108611/x86_64/build.log", size_bytes: 284654128}
76
+ date_years:
77
+ key: start_time
78
+ rows_with_date: 20284
79
+ counts: {"2026": 20284}
80
+ media_type: {text/plain: 20284}
81
+ facets:
82
+ arch:
83
+ aarch64: 3566
84
+ i386: 308
85
+ i686: 1681
86
+ image/aarch64: 1345
87
+ image/ppc64le: 276
88
+ image/s390x: 238
89
+ image/x86_64: 1401
90
+ noarch: 876
91
+ ppc64le: 3504
92
+ s390x: 3350
93
+ x86_64: 3739
94
+ build_state: {COMPLETE: 17207, FAILED: 3077}
95
+ log_name:
96
+ build.log: 17638
97
+ dnf.librepo.log: 30
98
+ dnf.log: 30
99
+ dnf.rpm.log: 30
100
+ dnf5.log: 401
101
+ hw_info.log: 431
102
+ mock_config.log: 431
103
+ mock_output.log: 431
104
+ root.log: 431
105
+ state.log: 431
106
+ task_state: {CANCELED: 207, CLOSED: 685, FAILED: 2185}
107
+ artifact: null
108
+ source_origin: null
109
+ parse_cost: null
110
+ parse_quirks: null
111
+ structure:
112
+ self_describing: null
113
+ integrity_builtin: null
114
+ has_reference_output: null
115
+ invariants: null
116
+ tools: null
117
+ ships:
118
+ problem_statement: {present: null}
119
+ reference_solution: {share: null}
120
+ tests: {present: null}
121
+ env_definition: {present: null}
122
+ known_answer: {present: null}
123
+ generator: {present: null}
124
+ upstream_labels: {present: null}
125
+ class: null
126
+ subject:
127
+ domain_span: null
128
+ shape_span: null
129
+ skill_family_span: null
130
+ ont_domain: null
131
+ tech_vocabulary: null
132
+ risk:
133
+ contamination: null
134
+ overlap_check: null
135
+ pii: null
136
+ pii_evidence: null
137
+ malware: null
138
+ ship_bytes: unknown
139
+ subset:
140
+ rule: "take all rows; licence, size and version are labels, not filters"
141
+ rows_in_subset: 20284
142
+ last_built: null
143
+ access:
144
+ method: null
145
+ verify: null
146
+ hosts: {"https://kojipkgs.fedoraproject.org": 20284}
147
+ uri_example: "https://kojipkgs.fedoraproject.org/packages/ddccontrol/3.3.0/2.fc45/data/logs/aarch64/build.log"
148
+ needs_auth: {"false": 20284}
149
+ credential: null
150
+ hash_source: {computed_on_fetch: 20284}
151
+ range_supported: null
152
+ reachable:
153
+ index_flag: {unknown: 20284}
154
+ index_check: null
155
+ our_sample: null
156
+ drift: null
157
+ sample: null
158
+ evidence:
159
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
160
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
161
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
162
+ class: {fill: unfilled, basis: "ships not judged"}
163
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
164
+ form.by_ext:
165
+ fill: auto
166
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
167
+ checked_at: "2026-10-03"
168
+ form.date_years:
169
+ fill: auto
170
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
171
+ checked_at: "2026-10-03"
172
+ form.over_cap:
173
+ fill: auto
174
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
175
+ checked_at: "2026-10-03"
176
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
177
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
178
+ form.size_quantiles:
179
+ fill: auto
180
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
181
+ checked_at: "2026-10-03"
182
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
184
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ provenance.licence_tiers:
186
+ fill: auto
187
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
188
+ checked_at: "2026-10-03"
189
+ provenance.spdx_top:
190
+ fill: auto
191
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
192
+ checked_at: "2026-10-03"
193
+ risk.contamination: {fill: unfilled, basis: "this source was not part of the benchmark overlap scan"}
194
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
195
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
196
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
197
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
198
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
199
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ snapshot.rev:
211
+ fill: auto
212
+ basis: "last commit listed in the publish receipt of this source"
213
+ checked_at: "2026-10-03"
214
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ subset:
225
+ fill: auto
226
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
227
+ checked_at: "2026-10-03"
228
+ unit.pointer_count:
229
+ fill: auto
230
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
231
+ checked_at: "2026-10-03"
232
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
234
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/freebsd-bugzilla.yaml ADDED
@@ -0,0 +1,321 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: freebsd-bugzilla
8
+ parent: null
9
+ title: "FreeBSD Bugzilla bugs and attachments"
10
+ description: "All publicly visible bugs and public attachments on the FreeBSD Bugzilla (bugs.freebsd.org), listed through the Bugzilla REST API in bug number order. One pointer per bug (collection bug) and one per public attachment (collection attachment). Bugs cover the FreeBSD base system, ports and packages, and documentation; attachments are mostly port patches and new port shar archives, build logs, kernel panic messages and crash dump summaries, dmesg output and configuration files. For a bug the hash covers its comment list as returned by the API; an attachment is retrieved through the REST API, which returns its content base64 encoded. Product, component, version, platform, operating system, status, severity, keywords, attachment media type and an attachment category are in source_meta."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Copyright of the contributors; no license stated"
20
+ licence_url: "https://www.freebsd.org/privacy/"
21
+ licence_statement: "Bugs, comments and attachments remain the copyright of their submitters, and Bugzilla pages state no license. The FreeBSD privacy policy reads: \"We store information such as your email address, name, and organization or company ('Personally Identifiable Information') only if you decide to send us such information by submitting a FreeBSD bug or problem report ('PR'), subscribing to one of the FreeBSD mailing lists, posting content to our Site forums or sending a review to Phabricator. The FreeBSD Project allows unlimited distribution of that content. Information submitted in those reports and lists, including your Personally Identifiable Information, is considered public and will be accessible to anyone on the web.\" It does not state a license granted by submitters. Patches are likely BSD-2-Clause by FreeBSD convention (noted per row in source_meta.license_hint, not checked)."
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {unknown/unknown: 573793}
28
+ spdx_top: {"(none)": 573793}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 573793
34
+ units: {"API response": 573793}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 5672890870
38
+ rows_without_size: 0
39
+ collections:
40
+ distinct: 2
41
+ top: {attachment: 275132, bug: 298661}
42
+ sub_collections: null
43
+ batches: null
44
+ n_ext: 1079
45
+ usable_unit: null
46
+ form:
47
+ by_ext:
48
+ json: 298667
49
+ diff: 125421
50
+ patch: 74103
51
+ shar: 22483
52
+ txt: 14985
53
+ "(none)": 10809
54
+ log: 9145
55
+ gz: 2539
56
+ png: 1591
57
+ c: 1583
58
+ sh: 746
59
+ xz: 654
60
+ jpg: 606
61
+ bz2: 579
62
+ tgz: 450
63
+ p7s: 391
64
+ zip: 371
65
+ "1": 356
66
+ conf: 337
67
+ dat: 317
68
+ "(1059 other types)": 7660
69
+ bytes_p50: 1475
70
+ bytes_p95: 28934
71
+ bytes_max: 3620819
72
+ size_quantiles:
73
+ max: 3620819
74
+ min: 0
75
+ p25: 755
76
+ p50: 1475
77
+ p75: 3585
78
+ p90: 11388
79
+ p95: 28934
80
+ p99: 191748
81
+ p99.9: 806131
82
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
83
+ by_ext_size:
84
+ - {ext: json, rows: 298667, bytes: 873850902, p50: 1158, p95: 9792, max: 1113186}
85
+ - {ext: diff, rows: 125421, bytes: 966032033, p50: 1411, p95: 24736, max: 2869455}
86
+ - {ext: patch, rows: 74103, bytes: 753017050, p50: 1791, p95: 37496, max: 1234761}
87
+ - {ext: shar, rows: 22483, bytes: 283176065, p50: 3546, p95: 41488, max: 3304059}
88
+ - {ext: txt, rows: 14985, bytes: 329326633, p50: 2358, p95: 90885, max: 3620819}
89
+ - {ext: "(none)", rows: 10809, bytes: 160849905, p50: 1895, p95: 60930, max: 1144845}
90
+ - {ext: log, rows: 9145, bytes: 1046035797, p50: 36838, p95: 551791, max: 3574014}
91
+ - {ext: gz, rows: 2539, bytes: 151392435, p50: 9470, p95: 271057, max: 1049144}
92
+ - {ext: png, rows: 1591, bytes: 272965483, p50: 76505, p95: 674693, max: 1011294}
93
+ - {ext: c, rows: 1583, bytes: 15451606, p50: 897, p95: 20017, max: 957969}
94
+ - {ext: sh, rows: 746, bytes: 5628417, p50: 2881, p95: 21872, max: 310765}
95
+ - {ext: xz, rows: 654, bytes: 107662568, p50: 66832, p95: 717848, max: 1000252}
96
+ - {ext: jpg, rows: 606, bytes: 193827994, p50: 213473, p95: 927758, max: 1020131}
97
+ - {ext: bz2, rows: 579, bytes: 42916771, p50: 21288, p95: 286872, max: 1010338}
98
+ - {ext: tgz, rows: 450, bytes: 22307974, p50: 6895, p95: 278090, max: 886371}
99
+ - {ext: p7s, rows: 391, bytes: 1411613, p50: 3393, p95: 6010, max: 6464}
100
+ - {ext: zip, rows: 371, bytes: 40237756, p50: 37491, p95: 454628, max: 1009550}
101
+ - {ext: "1", rows: 356, bytes: 20185680, p50: 2817, p95: 295977, max: 880247}
102
+ - {ext: conf, rows: 337, bytes: 1180369, p50: 939, p95: 11444, max: 145928}
103
+ - {ext: dat, rows: 317, bytes: 764965, p50: 215, p95: 4818, max: 204800}
104
+ over_cap:
105
+ cap: 61944279
106
+ rows_over: 0
107
+ bytes_over: 0
108
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
109
+ largest:
110
+ - {pointer_id: "freebsd-bugzilla:attachment:89320", size_bytes: 3620819}
111
+ - {pointer_id: "freebsd-bugzilla:attachment:17305", size_bytes: 3574014}
112
+ - {pointer_id: "freebsd-bugzilla:attachment:95387", size_bytes: 3304059}
113
+ - {pointer_id: "freebsd-bugzilla:attachment:70541", size_bytes: 2869455}
114
+ - {pointer_id: "freebsd-bugzilla:attachment:86273", size_bytes: 1263455}
115
+ date_years: null
116
+ media_type:
117
+ application/json: 298665
118
+ application/octet-stream: 2016
119
+ application/x-shar: 1331
120
+ image/jpeg: 697
121
+ image/png: 1584
122
+ text/plain: 255181
123
+ "text/plain; charset=us-ascii": 1549
124
+ text/x-diff: 762
125
+ text/x-log: 1914
126
+ text/x-patch: 1049
127
+ facets:
128
+ category: {config: 448, crash: 1133, log: 13099, other: 26850, patch: 233602}
129
+ component:
130
+ "Books & Articles": 9421
131
+ "Bug Tracker": 538
132
+ "Individual Port(s)": 225898
133
+ "Manual Pages": 976
134
+ "Ports Framework": 1477
135
+ Trash: 7161
136
+ Website: 684
137
+ amd64: 628
138
+ arm: 693
139
+ bhyve: 242
140
+ bin: 13276
141
+ conf: 2673
142
+ gnu: 540
143
+ i386: 2764
144
+ kern: 21966
145
+ misc: 4904
146
+ standards: 518
147
+ tests: 316
148
+ usb: 1393
149
+ wireless: 731
150
+ is_obsolete: {"false": 241751, "true": 33381}
151
+ is_patch: {"false": 65213, "true": 209919}
152
+ platform:
153
+ Any: 286385
154
+ amd64: 9191
155
+ arm: 748
156
+ arm64: 921
157
+ i386: 750
158
+ ia64: 4
159
+ mips: 64
160
+ pc98: 1
161
+ powerpc: 466
162
+ riscv: 95
163
+ sparc64: 36
164
+ product:
165
+ "Base System": 51430
166
+ Community: 12
167
+ Documentation: 11122
168
+ Other: 206
169
+ "Ports & Packages": 227509
170
+ Services: 8382
171
+ resolution:
172
+ DUPLICATE: 3135
173
+ FIXED: 264758
174
+ "Feedback Timeout": 2413
175
+ "No AI Content": 3
176
+ "Not A Bug": 2791
177
+ "Not Accepted": 1596
178
+ "Not Enough Information": 489
179
+ "Overcome By Events": 10660
180
+ "Please Upstream": 59
181
+ "Unable to Reproduce": 1259
182
+ "Works As Intended": 1996
183
+ severity: {"Affects Many People": 13124, "Affects Only Me": 257356, "Affects Some People": 28181}
184
+ status: {Closed: 289159, "In Progress": 574, New: 5142, Open: 3786}
185
+ artifact: null
186
+ source_origin: null
187
+ parse_cost: null
188
+ parse_quirks: null
189
+ structure:
190
+ self_describing: null
191
+ integrity_builtin: null
192
+ has_reference_output: null
193
+ invariants: null
194
+ tools: null
195
+ ships:
196
+ problem_statement: {present: null}
197
+ reference_solution: {share: null}
198
+ tests: {present: null}
199
+ env_definition: {present: null}
200
+ known_answer: {present: null}
201
+ generator: {present: null}
202
+ upstream_labels: {present: null}
203
+ class: null
204
+ subject:
205
+ domain_span: null
206
+ shape_span: null
207
+ skill_family_span: null
208
+ ont_domain: null
209
+ tech_vocabulary: null
210
+ risk:
211
+ contamination: none
212
+ overlap_check:
213
+ done: true
214
+ pointers_flagged: 0
215
+ pointers_removed: 0
216
+ by_class: {}
217
+ benches: {}
218
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
219
+ pii: null
220
+ pii_evidence: null
221
+ malware: null
222
+ ship_bytes: unknown
223
+ subset:
224
+ rule: "take all rows; licence, size and version are labels, not filters"
225
+ rows_in_subset: 573793
226
+ last_built: null
227
+ access:
228
+ method: "Bug rows and attachment rows: GET uri; the response is JSON."
229
+ verify: "Bug rows: take bugs.<bug number>.comments (the path is in source_meta.api_body_field; a list of objects with a single text key), serialize it with json.dumps(ensure_ascii=False), encode it as UTF-8 and compare size and sha256; the stored file is that JSON. Attachment rows: take attachments.<attachment number>.data (the path is in source_meta.api_body_field), decode it as base64 (source_meta.api_body_decode) and compare the size of the decoded bytes; the stored file is the decoded attachment. The source publishes no hash, so compute it on download."
230
+ hosts: {"https://bugs.freebsd.org": 573793}
231
+ uri_example: "https://bugs.freebsd.org/bugzilla/rest/bug/attachment/1"
232
+ needs_auth: {"false": 573793}
233
+ credential: null
234
+ hash_source: {computed: 298661, computed_on_fetch: 275132}
235
+ range_supported: null
236
+ reachable:
237
+ index_flag: {unknown: 573793}
238
+ index_check: null
239
+ our_sample: null
240
+ drift: null
241
+ sample: null
242
+ evidence:
243
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
244
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
245
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ class: {fill: unfilled, basis: "ships not judged"}
247
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ form.by_ext:
249
+ fill: auto
250
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
251
+ checked_at: "2026-10-03"
252
+ form.date_years:
253
+ fill: auto
254
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
255
+ checked_at: "2026-10-03"
256
+ form.over_cap:
257
+ fill: auto
258
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
259
+ checked_at: "2026-10-03"
260
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
261
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
262
+ form.size_quantiles:
263
+ fill: auto
264
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
265
+ checked_at: "2026-10-03"
266
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
268
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ provenance.licence_tiers:
270
+ fill: auto
271
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
272
+ checked_at: "2026-10-03"
273
+ provenance.spdx_top:
274
+ fill: auto
275
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
276
+ checked_at: "2026-10-03"
277
+ risk.contamination:
278
+ fill: auto
279
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
280
+ checked_at: "2026-10-03"
281
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
285
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
287
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
290
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
293
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
294
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
295
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
296
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
297
+ snapshot.rev:
298
+ fill: auto
299
+ basis: "last commit listed in the publish receipt of this source"
300
+ checked_at: "2026-10-03"
301
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
302
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
303
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
304
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
305
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
306
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
307
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
308
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
309
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
310
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
311
+ subset:
312
+ fill: auto
313
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
314
+ checked_at: "2026-10-03"
315
+ unit.pointer_count:
316
+ fill: auto
317
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
318
+ checked_at: "2026-10-03"
319
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
320
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
321
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/fuse-spreadsheets-common-crawl.yaml ADDED
@@ -0,0 +1,237 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: fuse-spreadsheets-common-crawl
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: null
20
+ licence_url: null
21
+ licence_statement: null
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {unknown/unknown: 249376}
28
+ spdx_top: {"(none)": 249376}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 249376
34
+ units: {"WARC record": 249376}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 7589191487
38
+ rows_without_size: 0
39
+ collections:
40
+ distinct: 10
41
+ top:
42
+ CC-MAIN-2013-20: 30202
43
+ CC-MAIN-2013-48: 23447
44
+ CC-MAIN-2014-10: 21867
45
+ CC-MAIN-2014-15: 19459
46
+ CC-MAIN-2014-23: 28244
47
+ CC-MAIN-2014-35: 23651
48
+ CC-MAIN-2014-41: 23221
49
+ CC-MAIN-2014-42: 37308
50
+ CC-MAIN-2014-49: 18454
51
+ CC-MAIN-2014-52: 23523
52
+ sub_collections: null
53
+ batches: null
54
+ n_ext: 3
55
+ usable_unit: null
56
+ form:
57
+ by_ext: {xls: 238673, xlsx: 10555, xltx: 148}
58
+ bytes_p50: 4295
59
+ bytes_p95: 176432
60
+ bytes_max: 1044831
61
+ size_quantiles:
62
+ max: 1044831
63
+ min: 1163
64
+ p25: 3516
65
+ p50: 4295
66
+ p75: 10049
67
+ p90: 75354
68
+ p95: 176432
69
+ p99: 400096
70
+ p99.9: 955594
71
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
72
+ by_ext_size:
73
+ - {ext: xls, rows: 238673, bytes: 6631353241, p50: 4007, p95: 169560, max: 1044831}
74
+ - {ext: xlsx, rows: 10555, bytes: 952920659, p50: 30923, p95: 458218, max: 1019038}
75
+ - {ext: xltx, rows: 148, bytes: 4917587, p50: 14125, p95: 147260, max: 379904}
76
+ over_cap:
77
+ cap: 61944279
78
+ rows_over: 0
79
+ bytes_over: 0
80
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
81
+ largest:
82
+ - pointer_id: "fuse-spreadsheets-common-crawl:CC-MAIN-2014-10:CC-MAIN-20140305060758-00012-ip-10-183-142-35.ec2.internal.warc.gz@443491682"
83
+ size_bytes: 1044831
84
+ - pointer_id: "fuse-spreadsheets-common-crawl:CC-MAIN-2014-10:CC-MAIN-20140305060726-00068-ip-10-183-142-35.ec2.internal.warc.gz@622028092"
85
+ size_bytes: 1044763
86
+ - pointer_id: "fuse-spreadsheets-common-crawl:CC-MAIN-2014-15:CC-MAIN-20140416005218-00563-ip-10-147-4-33.ec2.internal.warc.gz@381196161"
87
+ size_bytes: 1044411
88
+ - pointer_id: "fuse-spreadsheets-common-crawl:CC-MAIN-2014-15:CC-MAIN-20140416005219-00571-ip-10-147-4-33.ec2.internal.warc.gz@811531593"
89
+ size_bytes: 1043522
90
+ - pointer_id: "fuse-spreadsheets-common-crawl:CC-MAIN-2014-42:CC-MAIN-20141024030048-00033-ip-10-16-133-185.ec2.internal.warc.gz@513554130"
91
+ size_bytes: 1043375
92
+ date_years:
93
+ key: origin.captured_at
94
+ rows_with_date: 249376
95
+ counts: {"2013": 53649, "2014": 195727}
96
+ media_type:
97
+ application/vnd.ms-excel: 238673
98
+ application/vnd.openxmlformats-officedocument.spreadsheetml.sheet: 10555
99
+ application/vnd.openxmlformats-officedocument.spreadsheetml.template: 148
100
+ facets: null
101
+ artifact: null
102
+ source_origin: null
103
+ parse_cost: null
104
+ parse_quirks: null
105
+ structure:
106
+ self_describing: null
107
+ integrity_builtin: null
108
+ has_reference_output: null
109
+ invariants: null
110
+ tools: null
111
+ ships:
112
+ problem_statement: {present: null}
113
+ reference_solution: {share: null}
114
+ tests: {present: null}
115
+ env_definition: {present: null}
116
+ known_answer: {present: null}
117
+ generator: {present: null}
118
+ upstream_labels: {present: null}
119
+ class: null
120
+ subject:
121
+ domain_span: null
122
+ shape_span: null
123
+ skill_family_span: null
124
+ ont_domain: null
125
+ tech_vocabulary: null
126
+ risk:
127
+ contamination: none
128
+ overlap_check:
129
+ done: true
130
+ pointers_flagged: 0
131
+ pointers_removed: 0
132
+ by_class: {}
133
+ benches: {}
134
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
135
+ pii: null
136
+ pii_evidence: null
137
+ malware: null
138
+ ship_bytes: unknown
139
+ subset:
140
+ rule: "take all rows; licence, size and version are labels, not filters"
141
+ rows_in_subset: 249376
142
+ last_built: null
143
+ access:
144
+ method: null
145
+ verify: null
146
+ hosts: {"s3://commoncrawl": 249376}
147
+ uri_example: "s3://commoncrawl/crawl-data/CC-MAIN-2013-20/segments/1368696381630/warc/CC-MAIN-20130516092621-00000-ip-10-60-113-184.ec2.internal.warc.gz"
148
+ needs_auth: {"false": 249376}
149
+ credential: null
150
+ hash_source: {source_published: 249376}
151
+ range_supported: null
152
+ reachable:
153
+ index_flag: {"yes": 249376}
154
+ index_check: null
155
+ our_sample: null
156
+ drift: null
157
+ sample: null
158
+ evidence:
159
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
160
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
161
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
162
+ class: {fill: unfilled, basis: "ships not judged"}
163
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
164
+ form.by_ext:
165
+ fill: auto
166
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
167
+ checked_at: "2026-10-03"
168
+ form.date_years:
169
+ fill: auto
170
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
171
+ checked_at: "2026-10-03"
172
+ form.over_cap:
173
+ fill: auto
174
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
175
+ checked_at: "2026-10-03"
176
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
177
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
178
+ form.size_quantiles:
179
+ fill: auto
180
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
181
+ checked_at: "2026-10-03"
182
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
184
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ provenance.licence_tiers:
186
+ fill: auto
187
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
188
+ checked_at: "2026-10-03"
189
+ provenance.spdx_top:
190
+ fill: auto
191
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
192
+ checked_at: "2026-10-03"
193
+ risk.contamination:
194
+ fill: auto
195
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
196
+ checked_at: "2026-10-03"
197
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
198
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
199
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
201
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
212
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
213
+ snapshot.rev:
214
+ fill: auto
215
+ basis: "last commit listed in the publish receipt of this source"
216
+ checked_at: "2026-10-03"
217
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ subset:
228
+ fill: auto
229
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
230
+ checked_at: "2026-10-03"
231
+ unit.pointer_count:
232
+ fill: auto
233
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
234
+ checked_at: "2026-10-03"
235
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
237
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/fuse-spreadsheets.yaml ADDED
@@ -0,0 +1,216 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: fuse-spreadsheets
8
+ parent: null
9
+ title: "Fuse (Zenodo record 581678)"
10
+ description: "Files of Zenodo record 581678: Fuse (DOI 10.5281/zenodo.581678). One pointer per file."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: "9cb06b45f4359ffa848282f67ec89f0da37d3789", at: "2026-10-01", rev_candidates: 3}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: cc-by-4.0
17
+ licence_url: "https://zenodo.org/records/581678"
18
+ licence_statement: "License in the Zenodo record metadata: cc-by-4.0."
19
+ per_collection: null
20
+ training_ok: "yes"
21
+ redistribution_ok: "yes"
22
+ attribution: null
23
+ per_file: null
24
+ licence_tiers: {yes/yes: 1}
25
+ spdx_top: {"(none)": 1}
26
+ spdx_distinct: 1
27
+ licence_sample_check: null
28
+ unit:
29
+ pointer_level: null
30
+ pointer_count: 1
31
+ units: {file: 1}
32
+ state: null
33
+ expanded_count: null
34
+ bytes_total: 9392830118
35
+ rows_without_size: 0
36
+ collections: null
37
+ sub_collections: null
38
+ batches: null
39
+ n_ext: 1
40
+ usable_unit: null
41
+ form:
42
+ by_ext: {zip: 1}
43
+ bytes_p50: 9392830118
44
+ bytes_p95: 9392830118
45
+ bytes_max: 9392830118
46
+ size_quantiles:
47
+ max: 9392830118
48
+ min: 9392830118
49
+ p25: 9392830118
50
+ p50: 9392830118
51
+ p75: 9392830118
52
+ p90: 9392830118
53
+ p95: 9392830118
54
+ p99: 9392830118
55
+ p99.9: 9392830118
56
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
57
+ by_ext_size:
58
+ - {ext: zip, rows: 1, bytes: 9392830118, p50: 9392830118, p95: 9392830118, max: 9392830118}
59
+ over_cap:
60
+ cap: 61944279
61
+ rows_over: 1
62
+ bytes_over: 9392830118
63
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
64
+ largest:
65
+ - {pointer_id: "fuse-spreadsheets:zenodo-581678:fuse.zip", size_bytes: 9392830118}
66
+ date_years: null
67
+ media_type: {"(none)": 1}
68
+ facets: null
69
+ artifact: null
70
+ source_origin: null
71
+ parse_cost: null
72
+ parse_quirks: null
73
+ structure:
74
+ self_describing: null
75
+ integrity_builtin: null
76
+ has_reference_output: null
77
+ invariants: null
78
+ tools: null
79
+ ships:
80
+ problem_statement: {present: null}
81
+ reference_solution: {share: null}
82
+ tests: {present: null}
83
+ env_definition: {present: null}
84
+ known_answer: {present: null}
85
+ generator: {present: null}
86
+ upstream_labels: {present: null}
87
+ class: null
88
+ subject:
89
+ domain_span: null
90
+ shape_span: null
91
+ skill_family_span: null
92
+ ont_domain: null
93
+ tech_vocabulary: null
94
+ risk:
95
+ contamination: none
96
+ overlap_check:
97
+ done: true
98
+ pointers_flagged: 0
99
+ pointers_removed: 0
100
+ by_class: {}
101
+ benches: {}
102
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
103
+ pii: null
104
+ pii_evidence: null
105
+ malware: null
106
+ ship_bytes: "yes"
107
+ subset:
108
+ rule: "take all rows; licence, size and version are labels, not filters"
109
+ rows_in_subset: 1
110
+ last_built:
111
+ date: "2026-10-02"
112
+ rule: "older rule that filtered by licence tier"
113
+ rows: 1
114
+ rows_in_subset: 1
115
+ excluded_by_licence: 0
116
+ duplicates: 0
117
+ access:
118
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
119
+ verify: "Compare size_bytes and md5 (from the Zenodo API)."
120
+ hosts: {"https://zenodo.org": 1}
121
+ uri_example: "https://zenodo.org/records/581678/files/fuse.zip"
122
+ needs_auth: {"false": 1}
123
+ credential: null
124
+ hash_source: {source_published: 1}
125
+ range_supported: null
126
+ reachable:
127
+ index_flag: {"yes": 1}
128
+ index_check:
129
+ checked_at: "2026-10-01"
130
+ sample: null
131
+ checked: 1
132
+ ok: 1
133
+ by_status: {ok: 1}
134
+ our_sample: null
135
+ drift: null
136
+ sample: null
137
+ evidence:
138
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
139
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
140
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
141
+ class: {fill: unfilled, basis: "ships not judged"}
142
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
143
+ form.by_ext:
144
+ fill: auto
145
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
146
+ checked_at: "2026-10-03"
147
+ form.date_years:
148
+ fill: auto
149
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
150
+ checked_at: "2026-10-03"
151
+ form.over_cap:
152
+ fill: auto
153
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
154
+ checked_at: "2026-10-03"
155
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
156
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
157
+ form.size_quantiles:
158
+ fill: auto
159
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
160
+ checked_at: "2026-10-03"
161
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
162
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
163
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
164
+ provenance.licence_tiers:
165
+ fill: auto
166
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
167
+ checked_at: "2026-10-03"
168
+ provenance.spdx_top:
169
+ fill: auto
170
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
171
+ checked_at: "2026-10-03"
172
+ risk.contamination:
173
+ fill: auto
174
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
175
+ checked_at: "2026-10-03"
176
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
177
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
178
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
179
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
180
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
181
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
184
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
186
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
188
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
189
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
190
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
191
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
192
+ snapshot.rev:
193
+ fill: auto
194
+ basis: "last commit listed in the publish receipt of this source"
195
+ checked_at: "2026-10-03"
196
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
197
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
198
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
199
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ subset:
207
+ fill: auto
208
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
209
+ checked_at: "2026-10-03"
210
+ unit.pointer_count:
211
+ fill: auto
212
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
213
+ checked_at: "2026-10-03"
214
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
216
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/gcc-bugzilla.yaml ADDED
@@ -0,0 +1,286 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: gcc-bugzilla
8
+ parent: null
9
+ title: "GCC Bugzilla bugs and attachments"
10
+ description: "All publicly visible bugs and public attachments on GCC Bugzilla (gcc.gnu.org/bugzilla), listed through the Bugzilla REST API in bug number order. One pointer per bug (collection bug) and one per public attachment (collection attachment), such as preprocessed test cases, patches and build logs. For a bug the hash covers its comment list as returned by the API; product, component, version, status, severity, keywords and attachment counts are in source_meta."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-02"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "No license stated for bug content"
20
+ licence_url: "https://gcc.gnu.org/bugzilla/"
21
+ licence_statement: "GCC Bugzilla states no license for bugs, comments or attachments; they remain the copyright of their submitters. No terms of use were found for gcc.gnu.org or sourceware.org."
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {unknown/unknown: 186889}
28
+ spdx_top: {"(none)": 186889}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 186889
34
+ units: {"API response": 121659, file: 65230}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 4934458148
38
+ rows_without_size: 53
39
+ collections:
40
+ distinct: 2
41
+ top: {attachment: 65230, bug: 121659}
42
+ sub_collections: null
43
+ batches: null
44
+ n_ext: 758
45
+ usable_unit: null
46
+ form:
47
+ by_ext:
48
+ json: 121659
49
+ patch: 11637
50
+ c: 8127
51
+ txt: 5708
52
+ "(none)": 4088
53
+ diff: 4073
54
+ gz: 3878
55
+ i: 3804
56
+ cpp: 3284
57
+ bz2: 2903
58
+ ii: 2757
59
+ cc: 1735
60
+ f90: 1708
61
+ s: 1403
62
+ xz: 1272
63
+ log: 839
64
+ zip: 836
65
+ java: 609
66
+ out: 540
67
+ tgz: 496
68
+ "(738 other types)": 5533
69
+ bytes_p50: 3074
70
+ bytes_p95: 116195
71
+ bytes_max: 4574729
72
+ size_quantiles:
73
+ max: 4574729
74
+ min: 0
75
+ p25: 1369
76
+ p50: 3074
77
+ p75: 7156
78
+ p90: 24673
79
+ p95: 116195
80
+ p99: 626669
81
+ p99.9: 1011365
82
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
83
+ by_ext_size:
84
+ - {ext: json, rows: 121659, bytes: 708489084, p50: 3140, p95: 15778, max: 4058775}
85
+ - {ext: patch, rows: 11637, bytes: 98334502, p50: 2464, p95: 23808, max: 880664}
86
+ - {ext: c, rows: 8127, bytes: 179544022, p50: 508, p95: 143458, max: 1887095}
87
+ - {ext: txt, rows: 5708, bytes: 117729654, p50: 1651, p95: 63780, max: 1726008}
88
+ - {ext: "(none)", rows: 4088, bytes: 132834715, p50: 3447, p95: 133084, max: 1694599}
89
+ - {ext: diff, rows: 4073, bytes: 34190482, p50: 2341, p95: 26318, max: 833229}
90
+ - {ext: gz, rows: 3878, bytes: 679661563, p50: 99559, p95: 620170, max: 4574729}
91
+ - {ext: i, rows: 3804, bytes: 537172119, p50: 40227, p95: 640893, max: 3522835}
92
+ - {ext: cpp, rows: 3284, bytes: 45438299, p50: 618, p95: 13743, max: 1543149}
93
+ - {ext: bz2, rows: 2903, bytes: 372257948, p50: 68641, p95: 492657, max: 1562209}
94
+ - {ext: ii, rows: 2757, bytes: 772662058, p50: 71944, p95: 902149, max: 2128777}
95
+ - {ext: cc, rows: 1735, bytes: 88180593, p50: 745, p95: 466594, max: 1302543}
96
+ - {ext: f90, rows: 1708, bytes: 16490287, p50: 1084, p95: 37261, max: 1018360}
97
+ - {ext: s, rows: 1403, bytes: 61021145, p50: 4435, p95: 254987, max: 984632}
98
+ - {ext: xz, rows: 1272, bytes: 366052223, p50: 205620, p95: 844708, max: 1947824}
99
+ - {ext: log, rows: 839, bytes: 65249370, p50: 26655, p95: 390909, max: 1322114}
100
+ - {ext: zip, rows: 836, bytes: 136981496, p50: 25291, p95: 740997, max: 3327550}
101
+ - {ext: java, rows: 609, bytes: 1396037, p50: 768, p95: 9286, max: 71680}
102
+ - {ext: out, rows: 540, bytes: 103024719, p50: 10058, p95: 889219, max: 2014070}
103
+ - {ext: tgz, rows: 496, bytes: 60014349, p50: 11259, p95: 629658, max: 1703691}
104
+ over_cap:
105
+ cap: 61944279
106
+ rows_over: 0
107
+ bytes_over: 0
108
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
109
+ largest:
110
+ - {pointer_id: "gcc-bugzilla:attachment:2268", size_bytes: 4574729}
111
+ - {pointer_id: "gcc-bugzilla:attachment:666", size_bytes: 4192291}
112
+ - {pointer_id: "gcc-bugzilla:bug:15966", size_bytes: 4058775}
113
+ - {pointer_id: "gcc-bugzilla:bug:42560", size_bytes: 3840716}
114
+ - {pointer_id: "gcc-bugzilla:attachment:4011", size_bytes: 3522835}
115
+ date_years: null
116
+ media_type:
117
+ application/gzip: 1244
118
+ application/json: 121659
119
+ application/octet-stream: 6499
120
+ application/x-bzip: 605
121
+ application/x-gzip: 1276
122
+ "application/x-gzip ": 664
123
+ application/x-xz: 1123
124
+ text/plain: 44483
125
+ text/x-c++src: 811
126
+ text/x-csrc: 3298
127
+ facets:
128
+ is_obsolete: {"false": 60036, "true": 5194}
129
+ is_patch: {"false": 46238, "true": 18992}
130
+ product: {classpath: 1954, gcc: 119705}
131
+ resolution: {DUPLICATE: 14934, FIXED: 70089, INVALID: 11438, MOVED: 348, WONTFIX: 2907, WORKSFORME: 1772}
132
+ severity:
133
+ blocker: 624
134
+ critical: 3300
135
+ enhancement: 8907
136
+ major: 1316
137
+ minor: 2825
138
+ normal: 103932
139
+ trivial: 755
140
+ status:
141
+ ASSIGNED: 1642
142
+ CLOSED: 558
143
+ NEW: 10203
144
+ REOPENED: 153
145
+ RESOLVED: 100749
146
+ SUSPENDED: 206
147
+ UNCONFIRMED: 7497
148
+ VERIFIED: 181
149
+ WAITING: 470
150
+ artifact: null
151
+ source_origin: null
152
+ parse_cost: null
153
+ parse_quirks: null
154
+ structure:
155
+ self_describing: null
156
+ integrity_builtin: null
157
+ has_reference_output: null
158
+ invariants: null
159
+ tools: null
160
+ ships:
161
+ problem_statement: {present: null}
162
+ reference_solution: {share: null}
163
+ tests: {present: null}
164
+ env_definition: {present: null}
165
+ known_answer: {present: null}
166
+ generator: {present: null}
167
+ upstream_labels: {present: null}
168
+ class: null
169
+ subject:
170
+ domain_span: null
171
+ shape_span: null
172
+ skill_family_span: null
173
+ ont_domain: null
174
+ tech_vocabulary: null
175
+ risk:
176
+ contamination: none
177
+ overlap_check:
178
+ done: true
179
+ pointers_flagged: 0
180
+ pointers_removed: 0
181
+ by_class: {}
182
+ benches: {}
183
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
184
+ pii: null
185
+ pii_evidence: null
186
+ malware: null
187
+ ship_bytes: unknown
188
+ subset:
189
+ rule: "take all rows; licence, size and version are labels, not filters"
190
+ rows_in_subset: 186889
191
+ last_built: null
192
+ access:
193
+ method: "Bug rows: GET uri; the response is JSON. Attachment rows: GET uri and save the body."
194
+ verify: "Bug rows: take bugs.<bug number>.comments (the path is in source_meta.api_body_field; a list of objects with a single text key), serialize it with json.dumps(ensure_ascii=False), encode it as UTF-8 and compare size and sha256; the stored file is that JSON. Attachment rows: compare size; the source publishes no hash, so compute it on download."
195
+ hosts: {"https://gcc.gnu.org": 186889}
196
+ uri_example: "https://gcc.gnu.org/bugzilla/attachment.cgi?id=10"
197
+ needs_auth: {"false": 186889}
198
+ credential: null
199
+ hash_source: {computed: 121659, computed_on_fetch: 65230}
200
+ range_supported: null
201
+ reachable:
202
+ index_flag: {unknown: 186889}
203
+ index_check: null
204
+ our_sample: null
205
+ drift: null
206
+ sample: null
207
+ evidence:
208
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ class: {fill: unfilled, basis: "ships not judged"}
212
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
213
+ form.by_ext:
214
+ fill: auto
215
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
216
+ checked_at: "2026-10-03"
217
+ form.date_years:
218
+ fill: auto
219
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
220
+ checked_at: "2026-10-03"
221
+ form.over_cap:
222
+ fill: auto
223
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
224
+ checked_at: "2026-10-03"
225
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ form.size_quantiles:
228
+ fill: auto
229
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
230
+ checked_at: "2026-10-03"
231
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
233
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ provenance.licence_tiers:
235
+ fill: auto
236
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
237
+ checked_at: "2026-10-03"
238
+ provenance.spdx_top:
239
+ fill: auto
240
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
241
+ checked_at: "2026-10-03"
242
+ risk.contamination:
243
+ fill: auto
244
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
245
+ checked_at: "2026-10-03"
246
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
247
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
249
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
250
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
251
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
252
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
253
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
257
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
260
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
261
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
262
+ snapshot.rev:
263
+ fill: auto
264
+ basis: "last commit listed in the publish receipt of this source"
265
+ checked_at: "2026-10-03"
266
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
268
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
270
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
272
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
273
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
275
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ subset:
277
+ fill: auto
278
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
279
+ checked_at: "2026-10-03"
280
+ unit.pointer_count:
281
+ fill: auto
282
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
283
+ checked_at: "2026-10-03"
284
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
286
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/go-module-proxy-zips.yaml ADDED
@@ -0,0 +1,461 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: go-module-proxy-zips
8
+ parent: null
9
+ title: "Go module mirror (proxy.golang.org)"
10
+ description: "Module version zips on the Go module mirror proxy.golang.org, listed from the module index index.golang.org (the feed of module versions the mirror has cached, in the order it first cached them). One pointer per module version. Listing started from 2019-04-01."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-02"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per module (each module's own license)"
20
+ licence_url: "https://proxy.golang.org/"
21
+ licence_statement: "The mirror redistributes modules as published by their authors; each module carries its own license, and the operator states no license for the content. The mirror keeps only a temporary copy of a module whose license it cannot detect."
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: "Rows carry no per-module license; read the LICENSE file inside each zip."
27
+ licence_tiers: {unknown/unknown: 59120216}
28
+ spdx_top: {"(none)": 59120216}
29
+ spdx_distinct: 1
30
+ licence_sample_check: {sample_n: 10, declared: 10, unknown_or_missing: 0}
31
+ unit:
32
+ pointer_level: archive
33
+ pointer_count: 59120216
34
+ units: {file: 59120216}
35
+ state: packed
36
+ expanded_count: null
37
+ bytes_total: 0
38
+ rows_without_size: 59120216
39
+ collections:
40
+ distinct: 1000000
41
+ top:
42
+ buf.build/gen/go/bufbuild/buf/protocolbuffers/go: 32521
43
+ buf.build/gen/go/compassiot/api/protocolbuffers/go: 39202
44
+ buf.build/gen/go/compassiot/harvest/protocolbuffers/go: 36379
45
+ buf.build/gen/go/compassiot/model/protocolbuffers/go: 45472
46
+ "github.com/!homebrew/homebrew-core": 62793
47
+ github.com/arkadiyt/bounty-targets-data: 33379
48
+ github.com/confluentinc/kafka-rest: 33570
49
+ github.com/confluentinc/schema-registry: 31990
50
+ github.com/homebrew/homebrew-core: 78146
51
+ github.com/openshift/release: 42304
52
+ sub_collections: null
53
+ batches: null
54
+ n_ext: 1
55
+ usable_unit: "one zip of one version of one Go module (a whole repository or one module subdirectory of it); successive versions and pseudo-versions of a module overlap heavily"
56
+ form:
57
+ by_ext: {zip: 59120216}
58
+ bytes_p50: null
59
+ bytes_p95: null
60
+ bytes_max: null
61
+ size_quantiles: null
62
+ by_ext_size:
63
+ - {ext: zip, rows: 59120216, bytes: 0, p50: null, p95: null, max: null}
64
+ over_cap:
65
+ cap: 61944279
66
+ rows_over: 0
67
+ bytes_over: 0
68
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
69
+ largest: []
70
+ date_years:
71
+ key: index_timestamp
72
+ rows_with_date: 59120216
73
+ counts:
74
+ "2019": 543798
75
+ "2020": 1516928
76
+ "2021": 4951563
77
+ "2022": 5161748
78
+ "2023": 5859757
79
+ "2024": 10390277
80
+ "2025": 13886413
81
+ "2026": 16809732
82
+ media_type: {"(none)": 59120216}
83
+ facets:
84
+ is_pseudo: {"false": 27596711, "true": 31523505}
85
+ artifact: [source-code, configuration]
86
+ source_origin: [downloaded-source-archive]
87
+ parse_cost: needs_decompress
88
+ parse_quirks:
89
+ - "every member path starts with <module>@<version>/"
90
+ - "+incompatible versions and pre-module snapshots ship no go.mod"
91
+ - "a fork may keep the upstream module path in go.mod"
92
+ - "the proxy serves any tagged repository, not only Go code"
93
+ structure:
94
+ self_describing: true
95
+ integrity_builtin: [zip_crc, go_sum_hashes]
96
+ has_reference_output: false
97
+ invariants: [member_paths_start_with_module_at_version]
98
+ tools: [unzip, go]
99
+ ships:
100
+ problem_statement: {present: false}
101
+ reference_solution:
102
+ share: {k: 0, "n": 10}
103
+ tests:
104
+ present: true
105
+ coverage: "8 of 10 samples carry test files (_test.go files, rspec specs in #01); #00 (generated API types) and #03 (generated protobuf code) carry none"
106
+ runs: null
107
+ env_definition: {present: true, kind: dependency_manifest, runs: null}
108
+ known_answer: {present: false}
109
+ generator: {present: false}
110
+ upstream_labels: {present: false, fields: null}
111
+ class: manifest_not_run
112
+ subject:
113
+ domain_span: [D2.4, D1.2, D3.2]
114
+ shape_span: [S1_code_to_tests, S2_env_build_config]
115
+ skill_family_span: [skill.software, skill.systems]
116
+ ont_domain: [software-systems]
117
+ tech_vocabulary: [go, go.mod, go.sum, "module proxy", pseudo-version, "+incompatible", protobuf, kubernetes]
118
+ risk:
119
+ contamination: tb_overlap
120
+ overlap_check:
121
+ done: true
122
+ pointers_flagged: 6747
123
+ pointers_removed: 6747
124
+ by_class: {derivative_container: 6747}
125
+ benches:
126
+ agentslastexam: 25
127
+ androidworld: 13
128
+ browsecomp: 20
129
+ cybergym: 46
130
+ deepswe: 599
131
+ financeagent: 14
132
+ multiswe: 312
133
+ osworld: 40
134
+ swebench: 5602
135
+ terminalbench: 312
136
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
137
+ pii: present
138
+ pii_evidence:
139
+ sample_n: 10
140
+ files_with_person_names: 4
141
+ note: "account names in module paths; AUTHORS and MAINTAINERS member files; public attribution"
142
+ malware: null
143
+ ship_bytes: unknown
144
+ subset:
145
+ rule: "take all rows; licence, size and version are labels, not filters"
146
+ rows_in_subset: 59120216
147
+ last_built: null
148
+ access:
149
+ method: "Download each uri (a GOPROXY protocol address: uppercase letters in the module path and version are written as ! followed by the lowercase letter). For bulk downloads the operator recommends the /cached-only endpoint: insert /cached-only after the host."
150
+ verify: "The source publishes no zip size or file hash: size_bytes is null and hashes is empty; record the size and sha256 measured on retrieval. The h1: hash in go.sum and sum.golang.org is a hash of the unpacked file tree (golang.org/x/mod/sumdb/dirhash), not of the zip file."
151
+ hosts: {"https://proxy.golang.org": 59120216}
152
+ uri_example: "https://proxy.golang.org/031a-94-43-250-12.ngrok-free.app/core/src/maasagent/@v/v0.0.0-20230811145058-4087627750ce.zip"
153
+ needs_auth: {"false": 59120216}
154
+ credential: none
155
+ hash_source: {computed_on_fetch: 59120216}
156
+ range_supported: null
157
+ reachable:
158
+ index_flag: {unknown: 59120216}
159
+ index_check: null
160
+ our_sample: {checked_at: "2026-10-04", n_tried: 11, n_ok: 10, hash_checked: 10}
161
+ drift: "a module version, once in the proxy, is pinned by the checksum database and does not change; new versions and pseudo-versions arrive continuously"
162
+ sample:
163
+ method: "strata are file type times the size quartile of this source (q1 to q4 by its own p25, p50, p75; unk for rows without a size); half of the samples go one per stratum to the largest strata, the rest in proportion to stratum size (largest remainder); within a stratum the rows with the smallest sha256(seed:pointer_id) are taken, so the draw is reproducible from the index"
164
+ seed: 20261004
165
+ "n": 10
166
+ n_drawn: 10
167
+ n_read: 10
168
+ unread: []
169
+ rule_n: "10 when one file type holds at least 95 percent of rows, else 30"
170
+ size_quartiles: [null, null, null]
171
+ size_filter: "only files of at most 15000000 bytes are candidates"
172
+ strata_candidates: {"zip|unk": 59120216}
173
+ strata_total: 1
174
+ strata_taken: {"zip|unk": 59120216}
175
+ fetched_at: "2026-10-04"
176
+ replaced: 1
177
+ kept: false
178
+ kept_note: "sample bytes were deleted after reading; refetch by pointer_id and compare sha256"
179
+ files:
180
+ - "no": "#00"
181
+ pointer_id: "go-module-proxy-zips:github.com/grafana/grafana/apps/logsdrilldown:v0.0.0-20260508160113-dff033923bff.zip"
182
+ stratum: "zip|unk"
183
+ size_bytes: 87699
184
+ sha256: "8eed0f5b57152bdc1afe405421141509783a1583a69971bdebce88b27cfcdd92"
185
+ - "no": "#01"
186
+ pointer_id: "go-module-proxy-zips:github.com/sidekiq-scheduler/sidekiq-scheduler:v2.1.3+incompatible.zip"
187
+ stratum: "zip|unk"
188
+ size_bytes: 95181
189
+ sha256: "9a1f541abfe2471c4784c2053727e6fb0576109acca82db20ee223d0d7c2f796"
190
+ - "no": "#02"
191
+ pointer_id: "go-module-proxy-zips:github.com/open-telemetry/opentelemetry-collector-contrib/exporter/datasetexporter:v0.113.1-0.20241112234720-e38e45f97609.zip"
192
+ stratum: "zip|unk"
193
+ size_bytes: 50622
194
+ sha256: c8e6c51bb58a4407de65446011db8f5e56ed49804b6ec2e845e5d15719d8ead8
195
+ - "no": "#03"
196
+ pointer_id: "go-module-proxy-zips:buf.build/gen/go/crossplane/crossplane/protocolbuffers/go:v1.36.0-20260414191208-e968316118ac.1.zip"
197
+ stratum: "zip|unk"
198
+ size_bytes: 38296
199
+ sha256: "2e45a7c4f50c4e4db4782b96e6e0d0a59a9cd1e0d67470ad8693f676f91c1c4e"
200
+ - "no": "#04"
201
+ pointer_id: "go-module-proxy-zips:github.com/rancher-teja/rke:v1.3.15-rc1.zip"
202
+ stratum: "zip|unk"
203
+ size_bytes: 1109105
204
+ sha256: "0cd187941b117387bc14611f27dd97c29c120d30ce091d1800d53e846065000d"
205
+ - "no": "#05"
206
+ pointer_id: "go-module-proxy-zips:k8s.io/apimachinery:v0.0.0-20190212052820-5acd323cc893.zip"
207
+ stratum: "zip|unk"
208
+ size_bytes: 792651
209
+ sha256: "41ca4d7bf9633dc71047471e960839824ebfcadd40545559791609a71f6aefe1"
210
+ - "no": "#06"
211
+ pointer_id: "go-module-proxy-zips:github.com/brotherlogic/filecopier:v0.0.0-20250611225123-b61084a8245e.zip"
212
+ stratum: "zip|unk"
213
+ size_bytes: 35306
214
+ sha256: b2deede5e865017fea777c24f5ffe76467a60c8500d65ecaa38b820d4d4d6523
215
+ - "no": "#07"
216
+ pointer_id: "go-module-proxy-zips:k8s.io/kubernetes/staging/src/k8s.io/apiserver:v0.0.0-20240220195850-df1803f6015f.zip"
217
+ stratum: "zip|unk"
218
+ size_bytes: 2402389
219
+ sha256: "1eb3068d2f487134150ac45fbb74aa33bbf950adff108135acc1a581dc71d954"
220
+ - "no": "#08"
221
+ pointer_id: "go-module-proxy-zips:github.com/prometheus/prometheus:v0.0.0-20200601152650-06a6621b6ca8.zip"
222
+ stratum: "zip|unk"
223
+ size_bytes: 3786050
224
+ sha256: "392f1210189ec006b49665a646e0b78cd100cad8ccbdf8626eed4082f5465e66"
225
+ - "no": "#09"
226
+ pointer_id: "go-module-proxy-zips:github.com/containers/storage:v0.0.0-20160714093807-e7e9f753e811.zip"
227
+ stratum: "zip|unk"
228
+ size_bytes: 9190985
229
+ sha256: "772db80da067d146e0f89836fe050b5d400df2ff76a9a5943669afed59c32ad6"
230
+ evidence:
231
+ access.credential:
232
+ fill: judgment
233
+ basis: "all 10 fetched anonymously from proxy.golang.org from the mini egress"
234
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
235
+ sample_n: 10
236
+ checked_at: "2026-10-04"
237
+ access.drift:
238
+ fill: judgment
239
+ basis: "pointer versions are fixed strings; 7 of 10 are pseudo-versions"
240
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
241
+ sample_n: 10
242
+ checked_at: "2026-10-04"
243
+ access.range_supported: {fill: unfilled, basis: "no sample evidence given for this field"}
244
+ class:
245
+ fill: derived
246
+ basis: "rule over ships.tests, env_definition, known_answer, generator and structure; manifest_not_run when tests and a dependency manifest are present but the build was not run"
247
+ form.artifact:
248
+ fill: judgment
249
+ basis: "Go source files are most members in 8 of 10; #01 is Ruby source (a gem repository served as +incompatible); go.mod, go.sum, Makefiles and YAML are build and service configuration"
250
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
251
+ sample_n: 10
252
+ checked_at: "2026-10-04"
253
+ form.by_ext:
254
+ fill: auto
255
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
256
+ checked_at: "2026-10-03"
257
+ form.date_years:
258
+ fill: auto
259
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
260
+ checked_at: "2026-10-03"
261
+ form.over_cap:
262
+ fill: auto
263
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
264
+ checked_at: "2026-10-03"
265
+ form.parse_cost:
266
+ fill: judgment
267
+ basis: "zip archives, must be unpacked before reading"
268
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
269
+ sample_n: 10
270
+ checked_at: "2026-10-04"
271
+ form.parse_quirks:
272
+ fill: judgment
273
+ basis: "#01 and #09 have no go.mod (#01 v2.1.3+incompatible, #09 a 2016 snapshot); #04 is the zip of github.com/rancher-teja/rke but its go.mod says module github.com/rancher/rke; #01 is a Ruby gem (24 .rb files listed, rspec specs)"
274
+ files: ["#01", "#04", "#09"]
275
+ sample_n: 10
276
+ checked_at: "2026-10-04"
277
+ form.size_quantiles:
278
+ fill: auto
279
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
280
+ checked_at: "2026-10-03"
281
+ form.source_origin:
282
+ fill: judgment
283
+ basis: "module zips are fetched by URL from the module proxy; no git history ships in any zip"
284
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
285
+ sample_n: 10
286
+ checked_at: "2026-10-04"
287
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
288
+ provenance.licence_sample_check:
289
+ fill: judgment
290
+ basis: "a LICENSE file (MIT-LICENSE in #01) in all 10"
291
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
292
+ sample_n: 10
293
+ checked_at: "2026-10-04"
294
+ provenance.licence_tiers:
295
+ fill: auto
296
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
297
+ checked_at: "2026-10-03"
298
+ provenance.spdx_top:
299
+ fill: auto
300
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
301
+ checked_at: "2026-10-03"
302
+ risk.contamination:
303
+ fill: auto
304
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
305
+ checked_at: "2026-10-03"
306
+ risk.malware: {fill: unfilled, basis: "no sample evidence given for this field"}
307
+ risk.pii:
308
+ fill: judgment
309
+ basis: "module paths carry personal GitHub account names (#04 rancher-teja, #06 brotherlogic); #09 ships an AUTHORS file (56809 bytes) and #08 a MAINTAINERS file; the e-mail addresses seen in the digests are role or list mailboxes (#08 a Google group, #09 security@docker.com)"
310
+ files: ["#04", "#06", "#08", "#09"]
311
+ sample_n: 10
312
+ checked_at: "2026-10-04"
313
+ risk.pii_evidence:
314
+ fill: judgment
315
+ basis: "member lists and module paths"
316
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
317
+ sample_n: 10
318
+ checked_at: "2026-10-04"
319
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
320
+ ships.env_definition.kind:
321
+ fill: judgment
322
+ basis: "go.mod plus go.sum pin the dependency graph; 3 of 10 also ship a Dockerfile"
323
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
324
+ sample_n: 10
325
+ checked_at: "2026-10-04"
326
+ ships.env_definition.present:
327
+ fill: judgment
328
+ basis: "go.mod in 8 of 10 (#00, #02 to #08); Dockerfile in #04 (package/Dockerfile), #08 and #09"
329
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
330
+ sample_n: 10
331
+ checked_at: "2026-10-04"
332
+ ships.env_definition.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
333
+ ships.generator.present:
334
+ fill: judgment
335
+ basis: "nothing produces (input, answer) pairs"
336
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
337
+ sample_n: 10
338
+ checked_at: "2026-10-04"
339
+ ships.known_answer.present:
340
+ fill: judgment
341
+ basis: "no expected value file"
342
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
343
+ sample_n: 10
344
+ checked_at: "2026-10-04"
345
+ ships.problem_statement.present:
346
+ fill: judgment
347
+ basis: "no task text; READMEs only describe the module (#02 exporter config, #06 'Copies files amongst servers', #09 Docker README)"
348
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
349
+ sample_n: 10
350
+ checked_at: "2026-10-04"
351
+ ships.reference_solution.share:
352
+ fill: judgment
353
+ basis: "no sample carries a fix, patch or worked solution for a stated problem"
354
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
355
+ sample_n: 10
356
+ checked_at: "2026-10-04"
357
+ ships.tests.coverage:
358
+ fill: judgment
359
+ basis: "test-like path counts in the digests"
360
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
361
+ sample_n: 10
362
+ checked_at: "2026-10-04"
363
+ ships.tests.present:
364
+ fill: judgment
365
+ basis: "test-like paths in 8 of 10 samples: #01 16 (rspec), #02 9, #04 6, #05 120, #06 4, #07 402, #08 222, #09 338; #00 and #03 have none"
366
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
367
+ sample_n: 10
368
+ checked_at: "2026-10-04"
369
+ ships.tests.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
370
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no sample evidence given for this field"}
371
+ ships.upstream_labels.present:
372
+ fill: judgment
373
+ basis: "the module proxy attaches no category labels and go.mod has no keyword or category field"
374
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
375
+ sample_n: 10
376
+ checked_at: "2026-10-04"
377
+ snapshot.rev:
378
+ fill: auto
379
+ basis: "last commit listed in the publish receipt of this source"
380
+ checked_at: "2026-10-03"
381
+ structure.has_reference_output:
382
+ fill: judgment
383
+ basis: "no expected output file in any sample"
384
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
385
+ sample_n: 10
386
+ checked_at: "2026-10-04"
387
+ structure.integrity_builtin:
388
+ fill: judgment
389
+ basis: "zip CRC per member in all 10; go.sum lists hashes for dependencies in at least 6 of 10 (#00, #02, #04 to #07)"
390
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
391
+ sample_n: 10
392
+ checked_at: "2026-10-04"
393
+ structure.invariants:
394
+ fill: judgment
395
+ basis: "all member paths in the 10 digests start with the module path and version from the pointer"
396
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
397
+ sample_n: 10
398
+ checked_at: "2026-10-04"
399
+ structure.self_describing:
400
+ fill: judgment
401
+ basis: "go.mod names the module path, Go version and dependencies in 8 of 10 (#00, #02 to #08); file names and package clauses describe the rest"
402
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
403
+ sample_n: 10
404
+ checked_at: "2026-10-04"
405
+ structure.tools:
406
+ fill: judgment
407
+ basis: "unpack with unzip; build and test with the go command (#01 needs ruby instead)"
408
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
409
+ sample_n: 10
410
+ checked_at: "2026-10-04"
411
+ subject.domain_span:
412
+ fill: judgment
413
+ basis: "container and cluster tooling (#03 Crossplane, #04 RKE Kubernetes installer, #05 Kubernetes apimachinery, #07 Kubernetes apiserver, #09 Docker storage); distributed services (#05, #07, #08 Prometheus with its TSDB); logs and metrics pipelines (#00 Grafana logs app, #02 OpenTelemetry log exporter, #08 Prometheus); at most three main values"
414
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
415
+ sample_n: 10
416
+ checked_at: "2026-10-04"
417
+ subject.ont_domain:
418
+ fill: judgment
419
+ basis: "all 10 are software infrastructure modules"
420
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
421
+ sample_n: 10
422
+ checked_at: "2026-10-04"
423
+ subject.shape_span:
424
+ fill: judgment
425
+ basis: "source with tests in 8 of 10; go.mod, Makefiles and Dockerfiles for builds"
426
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
427
+ sample_n: 10
428
+ checked_at: "2026-10-04"
429
+ subject.skill_family_span:
430
+ fill: judgment
431
+ basis: "library and server code; cluster, container and monitoring systems"
432
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
433
+ sample_n: 10
434
+ checked_at: "2026-10-04"
435
+ subject.tech_vocabulary:
436
+ fill: judgment
437
+ basis: "terms seen in member names, go.mod heads and version strings"
438
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
439
+ sample_n: 10
440
+ checked_at: "2026-10-04"
441
+ subset:
442
+ fill: auto
443
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
444
+ checked_at: "2026-10-03"
445
+ unit.pointer_count:
446
+ fill: auto
447
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
448
+ checked_at: "2026-10-03"
449
+ unit.pointer_level:
450
+ fill: judgment
451
+ basis: "every sample is one module zip from the Go module proxy holding one module version's file tree"
452
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
453
+ sample_n: 10
454
+ checked_at: "2026-10-04"
455
+ unit.state: {fill: derived, basis: "packed when pointer_level is archive, else expanded"}
456
+ unit.usable_unit:
457
+ fill: judgment
458
+ basis: "10 samples are 10 different modules at one version each; 7 of 10 are pseudo-versions (v0.0.0-<time>-<commit>)"
459
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
460
+ sample_n: 10
461
+ checked_at: "2026-10-04"
cards/govdocs1-files.yaml ADDED
@@ -0,0 +1,1366 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: govdocs1-files
8
+ parent: govdocs1
9
+ title: "GovDocs1 files (Digital Corpora)"
10
+ description: "Single files inside the GovDocs1 zip archives (real files collected from U.S. .gov websites, mixed types), one pointer per file, each retrievable with HTTP Range requests on its archive."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Digital Corpora terms"
20
+ licence_url: "https://digitalcorpora.org/corpora/file-corpora/files/"
21
+ licence_statement: "The site states the files are \"freely available for research and may be (to the best of our knowledge) freely redistributed\"."
22
+ per_collection: null
23
+ training_ok: "yes"
24
+ redistribution_ok: "yes"
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {yes/yes: 986279}
28
+ spdx_top: {"(none)": 986279}
29
+ spdx_distinct: 1
30
+ licence_sample_check:
31
+ sample_n: 30
32
+ declared: 0
33
+ unknown_or_missing: 30
34
+ note: "U.S. government works; no licence text inside the files"
35
+ unit:
36
+ pointer_level: file
37
+ pointer_count: 986279
38
+ units: {"archive member": 986279}
39
+ state: expanded
40
+ expanded_count: null
41
+ bytes_total: 500971524282
42
+ rows_without_size: 0
43
+ collections: null
44
+ sub_collections: null
45
+ batches: null
46
+ n_ext: 64
47
+ usable_unit: "one file collected from a U.S. government web site"
48
+ form:
49
+ by_ext:
50
+ pdf: 231231
51
+ html: 214568
52
+ jpg: 109233
53
+ txt: 78285
54
+ doc: 76616
55
+ xls: 62635
56
+ ppt: 49701
57
+ gif: 36302
58
+ xml: 33458
59
+ ps: 22015
60
+ csv: 18360
61
+ gz: 13725
62
+ log: 9976
63
+ eps: 5191
64
+ unk: 5186
65
+ png: 4125
66
+ swf: 3476
67
+ dbase3: 2601
68
+ pps: 1619
69
+ rtf: 1125
70
+ "(44 other types)": 6851
71
+ bytes_p50: 56116
72
+ bytes_p95: 2067968
73
+ bytes_max: 1632715329
74
+ size_quantiles:
75
+ max: 1632715329
76
+ min: 7
77
+ p25: 19407
78
+ p50: 56116
79
+ p75: 239616
80
+ p90: 976910
81
+ p95: 2067968
82
+ p99: 7011840
83
+ p99.9: 29065890
84
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
85
+ by_ext_size:
86
+ - {ext: pdf, rows: 231231, bytes: 137311083639, p50: 158824, p95: 2827921, max: 72168407}
87
+ - {ext: html, rows: 214568, bytes: 13219078387, p50: 21371, p95: 174918, max: 66022085}
88
+ - {ext: jpg, rows: 109233, bytes: 37527942734, p50: 80800, p95: 1319982, max: 9995738}
89
+ - {ext: txt, rows: 78285, bytes: 50930598604, p50: 24711, p95: 1187074, max: 1632715329}
90
+ - {ext: doc, rows: 76616, bytes: 31055430066, p50: 78848, p95: 1408000, max: 195857408}
91
+ - {ext: xls, rows: 62635, bytes: 30064468211, p50: 49152, p95: 2085376, max: 216668672}
92
+ - {ext: ppt, rows: 49701, bytes: 128617944105, p50: 801792, p95: 9804288, max: 423724032}
93
+ - {ext: gif, rows: 36302, bytes: 3062115933, p50: 20035, p95: 606008, max: 10749167}
94
+ - {ext: xml, rows: 33458, bytes: 8689444541, p50: 15586, p95: 520643, max: 224018211}
95
+ - {ext: ps, rows: 22015, bytes: 28957952427, p50: 215036, p95: 5556770, max: 130596535}
96
+ - {ext: csv, rows: 18360, bytes: 3508273242, p50: 13286, p95: 873843, max: 200409368}
97
+ - {ext: gz, rows: 13725, bytes: 9090961313, p50: 196275, p95: 2636153, max: 85658628}
98
+ - {ext: log, rows: 9976, bytes: 4338049901, p50: 38747, p95: 1351052, max: 510825467}
99
+ - {ext: eps, rows: 5191, bytes: 2948460338, p50: 42941, p95: 1465343, max: 130583334}
100
+ - {ext: unk, rows: 5186, bytes: 1529005167, p50: 32767, p95: 775277, max: 104480768}
101
+ - {ext: png, rows: 4125, bytes: 1131291338, p50: 43307, p95: 1356439, max: 9405725}
102
+ - {ext: swf, rows: 3476, bytes: 1945615121, p50: 107502, p95: 2908020, max: 10487805}
103
+ - {ext: dbase3, rows: 2601, bytes: 20620961, p50: 5122, p95: 15384, max: 1854002}
104
+ - {ext: pps, rows: 1619, bytes: 3805430175, p50: 1259520, p95: 8071168, max: 43015168}
105
+ - {ext: rtf, rows: 1125, bytes: 490900270, p50: 52510, p95: 1184894, max: 52617394}
106
+ over_cap:
107
+ cap: 61944279
108
+ rows_over: 268
109
+ bytes_over: 39971610940
110
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
111
+ largest:
112
+ - {pointer_id: "govdocs1-files:zipfiles:186.zip!/186/186887.txt", size_bytes: 1632715329}
113
+ - {pointer_id: "govdocs1-files:zipfiles:094.zip!/094/094062.txt", size_bytes: 1509834320}
114
+ - {pointer_id: "govdocs1-files:zipfiles:458.zip!/458/458462.txt", size_bytes: 1436855446}
115
+ - {pointer_id: "govdocs1-files:zipfiles:118.zip!/118/118276.txt", size_bytes: 1426379502}
116
+ - {pointer_id: "govdocs1-files:zipfiles:028.zip!/028/028758.txt", size_bytes: 770758388}
117
+ date_years: null
118
+ media_type: {"(none)": 986279}
119
+ facets: null
120
+ artifact: [documents-text, structured-data, image]
121
+ source_origin: []
122
+ parse_cost: needs_tool
123
+ parse_quirks:
124
+ - "legacy OLE2 office formats from the 1990s and 2000s"
125
+ - "mixed encodings (ISO-8859 and UTF-8 HTML)"
126
+ - "scanned PDF text with OCR noise"
127
+ - "author and editor names left in office metadata"
128
+ structure:
129
+ self_describing: false
130
+ integrity_builtin: [zip_crc32]
131
+ has_reference_output: false
132
+ invariants: [zip_crc32_matches, pdf_parses]
133
+ tools: ["curl --range", "python zlib", pdftotext, libreoffice, file, exiftool]
134
+ ships:
135
+ problem_statement: {present: false}
136
+ reference_solution:
137
+ share: {k: 0, "n": 30}
138
+ tests: {present: false, coverage: none, runs: null}
139
+ env_definition: {present: false, kind: none, runs: null}
140
+ known_answer: {present: false}
141
+ generator: {present: false}
142
+ upstream_labels:
143
+ present: false
144
+ fields: []
145
+ class: raw
146
+ subject:
147
+ domain_span: [D3.5, D9.4]
148
+ shape_span: [S4_data_to_artifact]
149
+ skill_family_span: [skill.data, skill.media]
150
+ ont_domain: [language-documents, general]
151
+ tech_vocabulary: [OLE2, PDF, HTML, JPEG, EXIF, XLS, PPT, TSV, zip]
152
+ risk:
153
+ contamination: none
154
+ overlap_check:
155
+ done: true
156
+ pointers_flagged: 0
157
+ pointers_removed: 0
158
+ by_class: {}
159
+ benches: {}
160
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
161
+ pii: present
162
+ pii_evidence:
163
+ sample_n: 30
164
+ files_with_email: 1
165
+ files_with_named_individuals: 6
166
+ note: "mostly author fields in OLE2 metadata of government staff"
167
+ malware: null
168
+ ship_bytes: "yes"
169
+ subset:
170
+ rule: "take all rows; licence, size and version are labels, not filters"
171
+ rows_in_subset: 986279
172
+ last_built: null
173
+ access:
174
+ method: "Each pointer is one file inside a zip archive at uri; retrieve it with two HTTP Range requests, no full download. Request the 30-byte local file header at zip_entry.local_header_offset, read the file name length (bytes 26-27) and the extra field length (bytes 28-29), then request zip_entry.compressed_size bytes starting right after the header, name and extra field. Decompress by zip_entry.compression_method: 0 is stored as is, 8 is raw deflate. The archive itself is parent_pointer_id; member_path is the path inside it."
175
+ verify: "After decompression, compare the length with size_bytes and the CRC32 with hashes.crc32; a member that fails either check is not the listed file."
176
+ hosts: {"https://digitalcorpora.s3.amazonaws.com": 986279}
177
+ uri_example: "https://digitalcorpora.s3.amazonaws.com/corpora/files/govdocs1/zipfiles/000.zip"
178
+ needs_auth: {"false": 986279}
179
+ credential: none
180
+ hash_source: {source_published: 986279}
181
+ range_supported: null
182
+ reachable:
183
+ index_flag: {"yes": 986279}
184
+ index_check: null
185
+ our_sample: {checked_at: "2026-10-04", n_tried: 30, n_ok: 30, hash_checked: 30}
186
+ drift: "static corpus, unchanged since 2009; the only risk is the mirror going offline"
187
+ sample:
188
+ method: "strata are file type times the size quartile of this source (q1 to q4 by its own p25, p50, p75; unk for rows without a size); half of the samples go one per stratum to the largest strata, the rest in proportion to stratum size (largest remainder); within a stratum the rows with the smallest sha256(seed:pointer_id) are taken, so the draw is reproducible from the index"
189
+ seed: 20261004
190
+ "n": 30
191
+ n_drawn: 30
192
+ n_read: 30
193
+ unread: []
194
+ rule_n: "10 when one file type holds at least 95 percent of rows, else 30"
195
+ size_quartiles: [19407, 56116, 239616]
196
+ size_filter: "only files of at most 15000000 bytes are candidates"
197
+ strata_candidates:
198
+ "html|q1": 100347
199
+ "pdf|q4": 95262
200
+ "pdf|q3": 76643
201
+ "html|q2": 71811
202
+ "pdf|q2": 43695
203
+ "ppt|q4": 36274
204
+ "txt|q1": 35201
205
+ "html|q3": 34910
206
+ "jpg|q4": 32140
207
+ "jpg|q3": 31145
208
+ "doc|q3": 29581
209
+ "xls|q2": 28818
210
+ "doc|q2": 28336
211
+ "jpg|q2": 23472
212
+ "jpg|q1": 22476
213
+ "xml|q1": 18762
214
+ "xls|q3": 18327
215
+ "gif|q1": 17860
216
+ "doc|q4": 17755
217
+ "pdf|q1": 15572
218
+ "txt|q2": 15090
219
+ "txt|q3": 14328
220
+ "txt|q4": 13301
221
+ "csv|q1": 10563
222
+ "ppt|q3": 10339
223
+ "ps|q4": 10256
224
+ "xls|q4": 9993
225
+ "gif|q2": 8535
226
+ "xml|q2": 7596
227
+ "html|q4": 7472
228
+ "gif|q3": 7197
229
+ "gz|q4": 6279
230
+ "ps|q3": 6118
231
+ "xls|q1": 5229
232
+ "xml|q3": 4635
233
+ "csv|q2": 4382
234
+ "gz|q3": 3473
235
+ "log|q1": 3040
236
+ "ps|q1": 2907
237
+ "gif|q4": 2710
238
+ strata_total: 193
239
+ strata_taken:
240
+ "doc|q2": 28336
241
+ "doc|q3": 29581
242
+ "html|q1": 100347
243
+ "html|q2": 71811
244
+ "html|q3": 34910
245
+ "jpg|q1": 22476
246
+ "jpg|q2": 23472
247
+ "jpg|q3": 31145
248
+ "jpg|q4": 32140
249
+ "pdf|q2": 43695
250
+ "pdf|q3": 76643
251
+ "pdf|q4": 95262
252
+ "ppt|q4": 36274
253
+ "txt|q1": 35201
254
+ "xls|q2": 28818
255
+ fetched_at: "2026-10-04"
256
+ replaced: 0
257
+ kept: false
258
+ kept_note: "sample bytes were deleted after reading; refetch by pointer_id and compare sha256"
259
+ files:
260
+ - "no": "#00"
261
+ pointer_id: "govdocs1-files:zipfiles:095.zip!/095/095071.doc"
262
+ stratum: "doc|q2"
263
+ size_bytes: 29184
264
+ sha256: "1d1dbf27dca398968fd100784867f4e3e19b22cfe6c405779c74d30e940d8842"
265
+ - "no": "#01"
266
+ pointer_id: "govdocs1-files:zipfiles:942.zip!/942/942223.doc"
267
+ stratum: "doc|q2"
268
+ size_bytes: 24064
269
+ sha256: "8df1cd00e97174d02f2ff35dbbc49a70203512d262e03afc98edd134f0908968"
270
+ - "no": "#02"
271
+ pointer_id: "govdocs1-files:zipfiles:776.zip!/776/776505.doc"
272
+ stratum: "doc|q3"
273
+ size_bytes: 95744
274
+ sha256: "9d33bf1a60edcd78bd564cae3edf325e3b0f933a58b89b297ccfe863553ae1f1"
275
+ - "no": "#03"
276
+ pointer_id: "govdocs1-files:zipfiles:754.zip!/754/754533.doc"
277
+ stratum: "doc|q3"
278
+ size_bytes: 207872
279
+ sha256: "345d87adb4d274a6a97d1f9c51ead109948e3cee757b84d5898674fc556a8740"
280
+ - "no": "#04"
281
+ pointer_id: "govdocs1-files:zipfiles:345.zip!/345/345951.html"
282
+ stratum: "html|q1"
283
+ size_bytes: 15527
284
+ sha256: "4aea94ff4d3dfc40121da35178e57e4c24abfbfd13ef3aa9e0d91509082265a1"
285
+ - "no": "#05"
286
+ pointer_id: "govdocs1-files:zipfiles:476.zip!/476/476744.html"
287
+ stratum: "html|q1"
288
+ size_bytes: 9409
289
+ sha256: "28b10590e8d5dc1fe04679aea77d8861684487544041342daf8eacf489c9215e"
290
+ - "no": "#06"
291
+ pointer_id: "govdocs1-files:zipfiles:516.zip!/516/516143.html"
292
+ stratum: "html|q1"
293
+ size_bytes: 17153
294
+ sha256: c21158864ae81228e5883170603a85b908ad2e47eb2bfdf0ad07810bc9bc726c
295
+ - "no": "#07"
296
+ pointer_id: "govdocs1-files:zipfiles:586.zip!/586/586597.html"
297
+ stratum: "html|q2"
298
+ size_bytes: 29453
299
+ sha256: aa48f01da5705bd905a878a9f857cb98afe95a2fe7f8538ad81a53f1ad7b9839
300
+ - "no": "#08"
301
+ pointer_id: "govdocs1-files:zipfiles:676.zip!/676/676594.html"
302
+ stratum: "html|q2"
303
+ size_bytes: 31571
304
+ sha256: "920e6132aff7041e1af9db16fe06d34426027014871cf71cfa07116e363b133a"
305
+ - "no": "#09"
306
+ pointer_id: "govdocs1-files:zipfiles:487.zip!/487/487755.html"
307
+ stratum: "html|q3"
308
+ size_bytes: 80818
309
+ sha256: "21729eda9530b7602d21f0f3e7bdd3ffd4a479914476e05694e1de6c753c2e2f"
310
+ - "no": "#10"
311
+ pointer_id: "govdocs1-files:zipfiles:718.zip!/718/718149.html"
312
+ stratum: "html|q3"
313
+ size_bytes: 89268
314
+ sha256: "6f03e77118a98a8c563fa9ae98a93e0bcc04ab6b068cb22cca202670751278de"
315
+ - "no": "#11"
316
+ pointer_id: "govdocs1-files:zipfiles:700.zip!/700/700824.jpg"
317
+ stratum: "jpg|q1"
318
+ size_bytes: 2533
319
+ sha256: f504cef82226991874db5b16f016d9f96603711aba0490696a54ea33ae0b1c3b
320
+ - "no": "#12"
321
+ pointer_id: "govdocs1-files:zipfiles:064.zip!/064/064807.jpg"
322
+ stratum: "jpg|q2"
323
+ size_bytes: 42849
324
+ sha256: "2ad484f7b1736a1d3d5db764424806e07eab78cccf987183a780553e6be25160"
325
+ - "no": "#13"
326
+ pointer_id: "govdocs1-files:zipfiles:969.zip!/969/969644.jpg"
327
+ stratum: "jpg|q3"
328
+ size_bytes: 65186
329
+ sha256: a0e47d606f0818da7ec7818915ebf83f9bdf5baedfb63a57d7120c2b7aac299e
330
+ - "no": "#14"
331
+ pointer_id: "govdocs1-files:zipfiles:553.zip!/553/553392.jpg"
332
+ stratum: "jpg|q3"
333
+ size_bytes: 100141
334
+ sha256: "085a3ca3d16401213ea237367918bd6b2e088dad57fd633c101c2463445cb94e"
335
+ - "no": "#15"
336
+ pointer_id: "govdocs1-files:zipfiles:939.zip!/939/939081.jpg"
337
+ stratum: "jpg|q4"
338
+ size_bytes: 589918
339
+ sha256: d8c111fee5edd78e8c88709c8dc25ec03ff16ca8a53375a735cdf947f9c6fcab
340
+ - "no": "#16"
341
+ pointer_id: "govdocs1-files:zipfiles:812.zip!/812/812705.jpg"
342
+ stratum: "jpg|q4"
343
+ size_bytes: 2441751
344
+ sha256: a029cb6ee7255bb074e1b30447f865430f80f892f3fa76a2fe1f480b1994b971
345
+ - "no": "#17"
346
+ pointer_id: "govdocs1-files:zipfiles:961.zip!/961/961198.pdf"
347
+ stratum: "pdf|q2"
348
+ size_bytes: 30453
349
+ sha256: "54718ae223d01e315b0fb26d39acb313bc8c0be422546bf2bd54c54b9ec318df"
350
+ - "no": "#18"
351
+ pointer_id: "govdocs1-files:zipfiles:309.zip!/309/309827.pdf"
352
+ stratum: "pdf|q2"
353
+ size_bytes: 23817
354
+ sha256: "026829ee3a43fde33d696b8a327a0ef3c819642a4f60f9025cad6cc2f687c4e4"
355
+ - "no": "#19"
356
+ pointer_id: "govdocs1-files:zipfiles:703.zip!/703/703864.pdf"
357
+ stratum: "pdf|q3"
358
+ size_bytes: 155127
359
+ sha256: "11b7186c4c7f214b8f7ce8cc7281cf5ea5ceeb945a4d4a5e6ec6f08b40830cd8"
360
+ - "no": "#20"
361
+ pointer_id: "govdocs1-files:zipfiles:215.zip!/215/215474.pdf"
362
+ stratum: "pdf|q3"
363
+ size_bytes: 101412
364
+ sha256: "9542de9cd202a74a6f99d6b0e4f9c0397f7b2f689623635bf7d9149d5a41ae2d"
365
+ - "no": "#21"
366
+ pointer_id: "govdocs1-files:zipfiles:723.zip!/723/723118.pdf"
367
+ stratum: "pdf|q4"
368
+ size_bytes: 714473
369
+ sha256: d76ecd255a1ebf9bb300b1aeb42e63b2055826b5874dd8e992d1fad7633b808e
370
+ - "no": "#22"
371
+ pointer_id: "govdocs1-files:zipfiles:010.zip!/010/010954.pdf"
372
+ stratum: "pdf|q4"
373
+ size_bytes: 1275004
374
+ sha256: "7b821ea79df624fd8f7c4e0242e2b840815a394b0e60712c37523aeea1ff1513"
375
+ - "no": "#23"
376
+ pointer_id: "govdocs1-files:zipfiles:202.zip!/202/202209.pdf"
377
+ stratum: "pdf|q4"
378
+ size_bytes: 1529223
379
+ sha256: ec10c25d79bf66c4273d9487e00dd4004c8feb5d85ff1763991c5f273d847b26
380
+ - "no": "#24"
381
+ pointer_id: "govdocs1-files:zipfiles:115.zip!/115/115247.ppt"
382
+ stratum: "ppt|q4"
383
+ size_bytes: 3292672
384
+ sha256: "39e28be0f1cb268d8f63c13bc64f57c51d7219a1fe26495cebeedb300fe8a452"
385
+ - "no": "#25"
386
+ pointer_id: "govdocs1-files:zipfiles:814.zip!/814/814008.ppt"
387
+ stratum: "ppt|q4"
388
+ size_bytes: 7447040
389
+ sha256: "20591cce3fb228cde40b25cade9c941755e470975058d4c57971ba698f72f2ed"
390
+ - "no": "#26"
391
+ pointer_id: "govdocs1-files:zipfiles:456.zip!/456/456849.txt"
392
+ stratum: "txt|q1"
393
+ size_bytes: 681
394
+ sha256: "2c6eafe81b4ee17196919b9a3302137187640df7403306bc73411534a0290a50"
395
+ - "no": "#27"
396
+ pointer_id: "govdocs1-files:zipfiles:444.zip!/444/444638.txt"
397
+ stratum: "txt|q1"
398
+ size_bytes: 17368
399
+ sha256: aada80ef8b0b2acae88e021d1fac357a911352a2e86366933d3f0da5ef1ba4b9
400
+ - "no": "#28"
401
+ pointer_id: "govdocs1-files:zipfiles:780.zip!/780/780438.xls"
402
+ stratum: "xls|q2"
403
+ size_bytes: 26112
404
+ sha256: d6ef3e7c4a0cffbb350b480136931501b66dc92037720aa247f20918d735c675
405
+ - "no": "#29"
406
+ pointer_id: "govdocs1-files:zipfiles:866.zip!/866/866450.xls"
407
+ stratum: "xls|q2"
408
+ size_bytes: 22528
409
+ sha256: ddc90450dabeda1062b927a10656242cb1e35f06835ad10d389aef8a81ed8509
410
+ evidence:
411
+ access.credential:
412
+ fill: judgment
413
+ basis: "all 30 fetched by byte range from the public S3 bucket without a key"
414
+ files:
415
+ - "#00"
416
+ - "#01"
417
+ - "#02"
418
+ - "#03"
419
+ - "#04"
420
+ - "#05"
421
+ - "#06"
422
+ - "#07"
423
+ - "#08"
424
+ - "#09"
425
+ - "#10"
426
+ - "#11"
427
+ - "#12"
428
+ - "#13"
429
+ - "#14"
430
+ - "#15"
431
+ - "#16"
432
+ - "#17"
433
+ - "#18"
434
+ - "#19"
435
+ - "#20"
436
+ - "#21"
437
+ - "#22"
438
+ - "#23"
439
+ - "#24"
440
+ - "#25"
441
+ - "#26"
442
+ - "#27"
443
+ - "#28"
444
+ - "#29"
445
+ sample_n: 30
446
+ checked_at: "2026-10-04"
447
+ access.drift:
448
+ fill: judgment
449
+ basis: "zip files on digitalcorpora.s3.amazonaws.com with an alternate download host"
450
+ files:
451
+ - "#00"
452
+ - "#01"
453
+ - "#02"
454
+ - "#03"
455
+ - "#04"
456
+ - "#05"
457
+ - "#06"
458
+ - "#07"
459
+ - "#08"
460
+ - "#09"
461
+ - "#10"
462
+ - "#11"
463
+ - "#12"
464
+ - "#13"
465
+ - "#14"
466
+ - "#15"
467
+ - "#16"
468
+ - "#17"
469
+ - "#18"
470
+ - "#19"
471
+ - "#20"
472
+ - "#21"
473
+ - "#22"
474
+ - "#23"
475
+ - "#24"
476
+ - "#25"
477
+ - "#26"
478
+ - "#27"
479
+ - "#28"
480
+ - "#29"
481
+ sample_n: 30
482
+ checked_at: "2026-10-04"
483
+ access.range_supported: {fill: unfilled, basis: "no sample evidence given for this field"}
484
+ class:
485
+ fill: derived
486
+ basis: "rule over ships.tests, env_definition, known_answer, generator and structure; manifest_not_run when tests and a dependency manifest are present but the build was not run"
487
+ form.artifact:
488
+ fill: judgment
489
+ basis: "DOC, PPT, PDF and HTML documents (#00 to #10, #17 to #25); tab-separated auction bids and a two-column numeric table (#26, #27) and XLS sheets (#28, #29); JPEG photos and figures (#11 to #16)"
490
+ files:
491
+ - "#00"
492
+ - "#01"
493
+ - "#02"
494
+ - "#03"
495
+ - "#04"
496
+ - "#05"
497
+ - "#06"
498
+ - "#07"
499
+ - "#08"
500
+ - "#09"
501
+ - "#10"
502
+ - "#11"
503
+ - "#12"
504
+ - "#13"
505
+ - "#14"
506
+ - "#15"
507
+ - "#16"
508
+ - "#17"
509
+ - "#18"
510
+ - "#19"
511
+ - "#20"
512
+ - "#21"
513
+ - "#22"
514
+ - "#23"
515
+ - "#24"
516
+ - "#25"
517
+ - "#26"
518
+ - "#27"
519
+ - "#28"
520
+ - "#29"
521
+ sample_n: 30
522
+ checked_at: "2026-10-04"
523
+ form.by_ext:
524
+ fill: auto
525
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
526
+ checked_at: "2026-10-03"
527
+ form.date_years:
528
+ fill: auto
529
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
530
+ checked_at: "2026-10-03"
531
+ form.over_cap:
532
+ fill: auto
533
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
534
+ checked_at: "2026-10-03"
535
+ form.parse_cost:
536
+ fill: judgment
537
+ basis: "DOC, PPT and XLS are OLE2 binaries (file(1) says Composite Document File); PDFs need pdftotext"
538
+ files: ["#00", "#17", "#24", "#28"]
539
+ sample_n: 30
540
+ checked_at: "2026-10-04"
541
+ form.parse_quirks:
542
+ fill: judgment
543
+ basis: "Word 4.10 to 5.1 era files (#00 to #03, #24, #25, #28, #29); ISO-8859 HTML in #05 and #10; garbled first lines in the NASA scan #22 and the fax-like title of #02"
544
+ files: ["#00", "#02", "#05", "#10", "#22", "#28"]
545
+ sample_n: 30
546
+ checked_at: "2026-10-04"
547
+ form.size_quantiles:
548
+ fill: auto
549
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
550
+ checked_at: "2026-10-03"
551
+ form.source_origin:
552
+ fill: judgment
553
+ basis: "no source code in the sample"
554
+ files:
555
+ - "#00"
556
+ - "#01"
557
+ - "#02"
558
+ - "#03"
559
+ - "#04"
560
+ - "#05"
561
+ - "#06"
562
+ - "#07"
563
+ - "#08"
564
+ - "#09"
565
+ - "#10"
566
+ - "#11"
567
+ - "#12"
568
+ - "#13"
569
+ - "#14"
570
+ - "#15"
571
+ - "#16"
572
+ - "#17"
573
+ - "#18"
574
+ - "#19"
575
+ - "#20"
576
+ - "#21"
577
+ - "#22"
578
+ - "#23"
579
+ - "#24"
580
+ - "#25"
581
+ - "#26"
582
+ - "#27"
583
+ - "#28"
584
+ - "#29"
585
+ sample_n: 30
586
+ checked_at: "2026-10-04"
587
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
588
+ provenance.licence_sample_check:
589
+ fill: judgment
590
+ basis: "no licence statement in the digests"
591
+ files:
592
+ - "#00"
593
+ - "#01"
594
+ - "#02"
595
+ - "#03"
596
+ - "#04"
597
+ - "#05"
598
+ - "#06"
599
+ - "#07"
600
+ - "#08"
601
+ - "#09"
602
+ - "#10"
603
+ - "#11"
604
+ - "#12"
605
+ - "#13"
606
+ - "#14"
607
+ - "#15"
608
+ - "#16"
609
+ - "#17"
610
+ - "#18"
611
+ - "#19"
612
+ - "#20"
613
+ - "#21"
614
+ - "#22"
615
+ - "#23"
616
+ - "#24"
617
+ - "#25"
618
+ - "#26"
619
+ - "#27"
620
+ - "#28"
621
+ - "#29"
622
+ sample_n: 30
623
+ checked_at: "2026-10-04"
624
+ provenance.licence_tiers:
625
+ fill: auto
626
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
627
+ checked_at: "2026-10-03"
628
+ provenance.spdx_top:
629
+ fill: auto
630
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
631
+ checked_at: "2026-10-03"
632
+ risk.contamination:
633
+ fill: auto
634
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
635
+ checked_at: "2026-10-03"
636
+ risk.malware: {fill: unfilled, basis: "no sample evidence given for this field"}
637
+ risk.pii:
638
+ fill: judgment
639
+ basis: "office metadata names authors and editors (#00, #03, #24, #25, #29); #01 holds a staff email at mail.house.gov; #18 is a judicial council order; re-checked under the WP-T-44c definition (a user name on a platform counts as present)"
640
+ files: ["#00", "#01", "#03", "#18", "#24", "#25", "#29"]
641
+ sample_n: 30
642
+ checked_at: "2026-10-04"
643
+ risk.pii_evidence:
644
+ fill: judgment
645
+ basis: "file(1) metadata lines, email regex over the digests"
646
+ files: ["#00", "#01", "#03", "#24", "#25", "#29"]
647
+ sample_n: 30
648
+ checked_at: "2026-10-04"
649
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
650
+ ships.env_definition.kind:
651
+ fill: judgment
652
+ basis: "no environment definition"
653
+ files:
654
+ - "#00"
655
+ - "#01"
656
+ - "#02"
657
+ - "#03"
658
+ - "#04"
659
+ - "#05"
660
+ - "#06"
661
+ - "#07"
662
+ - "#08"
663
+ - "#09"
664
+ - "#10"
665
+ - "#11"
666
+ - "#12"
667
+ - "#13"
668
+ - "#14"
669
+ - "#15"
670
+ - "#16"
671
+ - "#17"
672
+ - "#18"
673
+ - "#19"
674
+ - "#20"
675
+ - "#21"
676
+ - "#22"
677
+ - "#23"
678
+ - "#24"
679
+ - "#25"
680
+ - "#26"
681
+ - "#27"
682
+ - "#28"
683
+ - "#29"
684
+ sample_n: 30
685
+ checked_at: "2026-10-04"
686
+ ships.env_definition.present:
687
+ fill: judgment
688
+ basis: "no environment definition"
689
+ files:
690
+ - "#00"
691
+ - "#01"
692
+ - "#02"
693
+ - "#03"
694
+ - "#04"
695
+ - "#05"
696
+ - "#06"
697
+ - "#07"
698
+ - "#08"
699
+ - "#09"
700
+ - "#10"
701
+ - "#11"
702
+ - "#12"
703
+ - "#13"
704
+ - "#14"
705
+ - "#15"
706
+ - "#16"
707
+ - "#17"
708
+ - "#18"
709
+ - "#19"
710
+ - "#20"
711
+ - "#21"
712
+ - "#22"
713
+ - "#23"
714
+ - "#24"
715
+ - "#25"
716
+ - "#26"
717
+ - "#27"
718
+ - "#28"
719
+ - "#29"
720
+ sample_n: 30
721
+ checked_at: "2026-10-04"
722
+ ships.env_definition.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
723
+ ships.generator.present:
724
+ fill: judgment
725
+ basis: "no generator"
726
+ files:
727
+ - "#00"
728
+ - "#01"
729
+ - "#02"
730
+ - "#03"
731
+ - "#04"
732
+ - "#05"
733
+ - "#06"
734
+ - "#07"
735
+ - "#08"
736
+ - "#09"
737
+ - "#10"
738
+ - "#11"
739
+ - "#12"
740
+ - "#13"
741
+ - "#14"
742
+ - "#15"
743
+ - "#16"
744
+ - "#17"
745
+ - "#18"
746
+ - "#19"
747
+ - "#20"
748
+ - "#21"
749
+ - "#22"
750
+ - "#23"
751
+ - "#24"
752
+ - "#25"
753
+ - "#26"
754
+ - "#27"
755
+ - "#28"
756
+ - "#29"
757
+ sample_n: 30
758
+ checked_at: "2026-10-04"
759
+ ships.known_answer.present:
760
+ fill: judgment
761
+ basis: "no answer field inside the files"
762
+ files:
763
+ - "#00"
764
+ - "#01"
765
+ - "#02"
766
+ - "#03"
767
+ - "#04"
768
+ - "#05"
769
+ - "#06"
770
+ - "#07"
771
+ - "#08"
772
+ - "#09"
773
+ - "#10"
774
+ - "#11"
775
+ - "#12"
776
+ - "#13"
777
+ - "#14"
778
+ - "#15"
779
+ - "#16"
780
+ - "#17"
781
+ - "#18"
782
+ - "#19"
783
+ - "#20"
784
+ - "#21"
785
+ - "#22"
786
+ - "#23"
787
+ - "#24"
788
+ - "#25"
789
+ - "#26"
790
+ - "#27"
791
+ - "#28"
792
+ - "#29"
793
+ sample_n: 30
794
+ checked_at: "2026-10-04"
795
+ ships.problem_statement.present:
796
+ fill: judgment
797
+ basis: "collected files, not tasks"
798
+ files:
799
+ - "#00"
800
+ - "#01"
801
+ - "#02"
802
+ - "#03"
803
+ - "#04"
804
+ - "#05"
805
+ - "#06"
806
+ - "#07"
807
+ - "#08"
808
+ - "#09"
809
+ - "#10"
810
+ - "#11"
811
+ - "#12"
812
+ - "#13"
813
+ - "#14"
814
+ - "#15"
815
+ - "#16"
816
+ - "#17"
817
+ - "#18"
818
+ - "#19"
819
+ - "#20"
820
+ - "#21"
821
+ - "#22"
822
+ - "#23"
823
+ - "#24"
824
+ - "#25"
825
+ - "#26"
826
+ - "#27"
827
+ - "#28"
828
+ - "#29"
829
+ sample_n: 30
830
+ checked_at: "2026-10-04"
831
+ ships.reference_solution.share:
832
+ fill: judgment
833
+ basis: "0 of 30 sample files read: no sample file carries a fix, patch, solved answer or worked solution"
834
+ files: []
835
+ sample_n: 30
836
+ checked_at: "2026-10-04"
837
+ ships.tests.coverage:
838
+ fill: judgment
839
+ basis: "no tests"
840
+ files:
841
+ - "#00"
842
+ - "#01"
843
+ - "#02"
844
+ - "#03"
845
+ - "#04"
846
+ - "#05"
847
+ - "#06"
848
+ - "#07"
849
+ - "#08"
850
+ - "#09"
851
+ - "#10"
852
+ - "#11"
853
+ - "#12"
854
+ - "#13"
855
+ - "#14"
856
+ - "#15"
857
+ - "#16"
858
+ - "#17"
859
+ - "#18"
860
+ - "#19"
861
+ - "#20"
862
+ - "#21"
863
+ - "#22"
864
+ - "#23"
865
+ - "#24"
866
+ - "#25"
867
+ - "#26"
868
+ - "#27"
869
+ - "#28"
870
+ - "#29"
871
+ sample_n: 30
872
+ checked_at: "2026-10-04"
873
+ ships.tests.present:
874
+ fill: judgment
875
+ basis: "no tests"
876
+ files:
877
+ - "#00"
878
+ - "#01"
879
+ - "#02"
880
+ - "#03"
881
+ - "#04"
882
+ - "#05"
883
+ - "#06"
884
+ - "#07"
885
+ - "#08"
886
+ - "#09"
887
+ - "#10"
888
+ - "#11"
889
+ - "#12"
890
+ - "#13"
891
+ - "#14"
892
+ - "#15"
893
+ - "#16"
894
+ - "#17"
895
+ - "#18"
896
+ - "#19"
897
+ - "#20"
898
+ - "#21"
899
+ - "#22"
900
+ - "#23"
901
+ - "#24"
902
+ - "#25"
903
+ - "#26"
904
+ - "#27"
905
+ - "#28"
906
+ - "#29"
907
+ sample_n: 30
908
+ checked_at: "2026-10-04"
909
+ ships.tests.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
910
+ ships.upstream_labels.fields:
911
+ fill: judgment
912
+ basis: none
913
+ files:
914
+ - "#00"
915
+ - "#01"
916
+ - "#02"
917
+ - "#03"
918
+ - "#04"
919
+ - "#05"
920
+ - "#06"
921
+ - "#07"
922
+ - "#08"
923
+ - "#09"
924
+ - "#10"
925
+ - "#11"
926
+ - "#12"
927
+ - "#13"
928
+ - "#14"
929
+ - "#15"
930
+ - "#16"
931
+ - "#17"
932
+ - "#18"
933
+ - "#19"
934
+ - "#20"
935
+ - "#21"
936
+ - "#22"
937
+ - "#23"
938
+ - "#24"
939
+ - "#25"
940
+ - "#26"
941
+ - "#27"
942
+ - "#28"
943
+ - "#29"
944
+ sample_n: 30
945
+ checked_at: "2026-10-04"
946
+ ships.upstream_labels.present:
947
+ fill: judgment
948
+ basis: "the pointers carry only zip name, member path and CRC; no topic or type label beyond the extension"
949
+ files:
950
+ - "#00"
951
+ - "#01"
952
+ - "#02"
953
+ - "#03"
954
+ - "#04"
955
+ - "#05"
956
+ - "#06"
957
+ - "#07"
958
+ - "#08"
959
+ - "#09"
960
+ - "#10"
961
+ - "#11"
962
+ - "#12"
963
+ - "#13"
964
+ - "#14"
965
+ - "#15"
966
+ - "#16"
967
+ - "#17"
968
+ - "#18"
969
+ - "#19"
970
+ - "#20"
971
+ - "#21"
972
+ - "#22"
973
+ - "#23"
974
+ - "#24"
975
+ - "#25"
976
+ - "#26"
977
+ - "#27"
978
+ - "#28"
979
+ - "#29"
980
+ sample_n: 30
981
+ checked_at: "2026-10-04"
982
+ snapshot.rev:
983
+ fill: auto
984
+ basis: "last commit listed in the publish receipt of this source"
985
+ checked_at: "2026-10-03"
986
+ structure.has_reference_output:
987
+ fill: judgment
988
+ basis: "no expected output file"
989
+ files:
990
+ - "#00"
991
+ - "#01"
992
+ - "#02"
993
+ - "#03"
994
+ - "#04"
995
+ - "#05"
996
+ - "#06"
997
+ - "#07"
998
+ - "#08"
999
+ - "#09"
1000
+ - "#10"
1001
+ - "#11"
1002
+ - "#12"
1003
+ - "#13"
1004
+ - "#14"
1005
+ - "#15"
1006
+ - "#16"
1007
+ - "#17"
1008
+ - "#18"
1009
+ - "#19"
1010
+ - "#20"
1011
+ - "#21"
1012
+ - "#22"
1013
+ - "#23"
1014
+ - "#24"
1015
+ - "#25"
1016
+ - "#26"
1017
+ - "#27"
1018
+ - "#28"
1019
+ - "#29"
1020
+ sample_n: 30
1021
+ checked_at: "2026-10-04"
1022
+ structure.integrity_builtin:
1023
+ fill: judgment
1024
+ basis: "every row carries the member CRC32 published in the zip directory"
1025
+ files:
1026
+ - "#00"
1027
+ - "#01"
1028
+ - "#02"
1029
+ - "#03"
1030
+ - "#04"
1031
+ - "#05"
1032
+ - "#06"
1033
+ - "#07"
1034
+ - "#08"
1035
+ - "#09"
1036
+ - "#10"
1037
+ - "#11"
1038
+ - "#12"
1039
+ - "#13"
1040
+ - "#14"
1041
+ - "#15"
1042
+ - "#16"
1043
+ - "#17"
1044
+ - "#18"
1045
+ - "#19"
1046
+ - "#20"
1047
+ - "#21"
1048
+ - "#22"
1049
+ - "#23"
1050
+ - "#24"
1051
+ - "#25"
1052
+ - "#26"
1053
+ - "#27"
1054
+ - "#28"
1055
+ - "#29"
1056
+ sample_n: 30
1057
+ checked_at: "2026-10-04"
1058
+ structure.invariants:
1059
+ fill: judgment
1060
+ basis: "all 30 members passed the CRC check on fetch; all 7 PDFs open in pdfinfo"
1061
+ files: ["#17", "#18", "#19", "#20", "#21", "#22", "#23"]
1062
+ sample_n: 30
1063
+ checked_at: "2026-10-04"
1064
+ structure.self_describing:
1065
+ fill: judgment
1066
+ basis: "office and PDF files carry no schema; only #26 has a header row"
1067
+ files: ["#26"]
1068
+ sample_n: 30
1069
+ checked_at: "2026-10-04"
1070
+ structure.tools:
1071
+ fill: judgment
1072
+ basis: "the tools that opened the samples"
1073
+ files:
1074
+ - "#00"
1075
+ - "#01"
1076
+ - "#02"
1077
+ - "#03"
1078
+ - "#04"
1079
+ - "#05"
1080
+ - "#06"
1081
+ - "#07"
1082
+ - "#08"
1083
+ - "#09"
1084
+ - "#10"
1085
+ - "#11"
1086
+ - "#12"
1087
+ - "#13"
1088
+ - "#14"
1089
+ - "#15"
1090
+ - "#16"
1091
+ - "#17"
1092
+ - "#18"
1093
+ - "#19"
1094
+ - "#20"
1095
+ - "#21"
1096
+ - "#22"
1097
+ - "#23"
1098
+ - "#24"
1099
+ - "#25"
1100
+ - "#26"
1101
+ - "#27"
1102
+ - "#28"
1103
+ - "#29"
1104
+ sample_n: 30
1105
+ checked_at: "2026-10-04"
1106
+ subject.domain_span:
1107
+ fill: judgment
1108
+ basis: "document and spreadsheet extraction across agencies (#00 to #10, #17 to #29); photos and figures (#11 to #16); trimmed to at most three main values (WP-T-44c): kept the values most samples show, dropped a parent code when its subcode is kept"
1109
+ files:
1110
+ - "#00"
1111
+ - "#01"
1112
+ - "#02"
1113
+ - "#03"
1114
+ - "#04"
1115
+ - "#05"
1116
+ - "#06"
1117
+ - "#07"
1118
+ - "#08"
1119
+ - "#09"
1120
+ - "#10"
1121
+ - "#11"
1122
+ - "#12"
1123
+ - "#13"
1124
+ - "#14"
1125
+ - "#15"
1126
+ - "#16"
1127
+ - "#17"
1128
+ - "#18"
1129
+ - "#19"
1130
+ - "#20"
1131
+ - "#21"
1132
+ - "#22"
1133
+ - "#23"
1134
+ - "#24"
1135
+ - "#25"
1136
+ - "#26"
1137
+ - "#27"
1138
+ - "#28"
1139
+ - "#29"
1140
+ sample_n: 30
1141
+ checked_at: "2026-10-04"
1142
+ subject.ont_domain:
1143
+ fill: judgment
1144
+ basis: "government documents on health (#04, #09, #23), legislation (#01, #05), audit (#19), space (#08, #22, #24), science (#20, #27)"
1145
+ files:
1146
+ - "#00"
1147
+ - "#01"
1148
+ - "#02"
1149
+ - "#03"
1150
+ - "#04"
1151
+ - "#05"
1152
+ - "#06"
1153
+ - "#07"
1154
+ - "#08"
1155
+ - "#09"
1156
+ - "#10"
1157
+ - "#11"
1158
+ - "#12"
1159
+ - "#13"
1160
+ - "#14"
1161
+ - "#15"
1162
+ - "#16"
1163
+ - "#17"
1164
+ - "#18"
1165
+ - "#19"
1166
+ - "#20"
1167
+ - "#21"
1168
+ - "#22"
1169
+ - "#23"
1170
+ - "#24"
1171
+ - "#25"
1172
+ - "#26"
1173
+ - "#27"
1174
+ - "#28"
1175
+ - "#29"
1176
+ sample_n: 30
1177
+ checked_at: "2026-10-04"
1178
+ subject.shape_span:
1179
+ fill: judgment
1180
+ basis: "files to extracted text, tables and metadata"
1181
+ files:
1182
+ - "#00"
1183
+ - "#01"
1184
+ - "#02"
1185
+ - "#03"
1186
+ - "#04"
1187
+ - "#05"
1188
+ - "#06"
1189
+ - "#07"
1190
+ - "#08"
1191
+ - "#09"
1192
+ - "#10"
1193
+ - "#11"
1194
+ - "#12"
1195
+ - "#13"
1196
+ - "#14"
1197
+ - "#15"
1198
+ - "#16"
1199
+ - "#17"
1200
+ - "#18"
1201
+ - "#19"
1202
+ - "#20"
1203
+ - "#21"
1204
+ - "#22"
1205
+ - "#23"
1206
+ - "#24"
1207
+ - "#25"
1208
+ - "#26"
1209
+ - "#27"
1210
+ - "#28"
1211
+ - "#29"
1212
+ sample_n: 30
1213
+ checked_at: "2026-10-04"
1214
+ subject.skill_family_span:
1215
+ fill: judgment
1216
+ basis: "office and PDF extraction; image metadata and OCR"
1217
+ files:
1218
+ - "#00"
1219
+ - "#01"
1220
+ - "#02"
1221
+ - "#03"
1222
+ - "#04"
1223
+ - "#05"
1224
+ - "#06"
1225
+ - "#07"
1226
+ - "#08"
1227
+ - "#09"
1228
+ - "#10"
1229
+ - "#11"
1230
+ - "#12"
1231
+ - "#13"
1232
+ - "#14"
1233
+ - "#15"
1234
+ - "#16"
1235
+ - "#17"
1236
+ - "#18"
1237
+ - "#19"
1238
+ - "#20"
1239
+ - "#21"
1240
+ - "#22"
1241
+ - "#23"
1242
+ - "#24"
1243
+ - "#25"
1244
+ - "#26"
1245
+ - "#27"
1246
+ - "#28"
1247
+ - "#29"
1248
+ sample_n: 30
1249
+ checked_at: "2026-10-04"
1250
+ subject.tech_vocabulary:
1251
+ fill: judgment
1252
+ basis: "formats seen in the sample"
1253
+ files:
1254
+ - "#00"
1255
+ - "#01"
1256
+ - "#02"
1257
+ - "#03"
1258
+ - "#04"
1259
+ - "#05"
1260
+ - "#06"
1261
+ - "#07"
1262
+ - "#08"
1263
+ - "#09"
1264
+ - "#10"
1265
+ - "#11"
1266
+ - "#12"
1267
+ - "#13"
1268
+ - "#14"
1269
+ - "#15"
1270
+ - "#16"
1271
+ - "#17"
1272
+ - "#18"
1273
+ - "#19"
1274
+ - "#20"
1275
+ - "#21"
1276
+ - "#22"
1277
+ - "#23"
1278
+ - "#24"
1279
+ - "#25"
1280
+ - "#26"
1281
+ - "#27"
1282
+ - "#28"
1283
+ - "#29"
1284
+ sample_n: 30
1285
+ checked_at: "2026-10-04"
1286
+ subset:
1287
+ fill: auto
1288
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
1289
+ checked_at: "2026-10-03"
1290
+ unit.pointer_count:
1291
+ fill: auto
1292
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
1293
+ checked_at: "2026-10-03"
1294
+ unit.pointer_level:
1295
+ fill: judgment
1296
+ basis: "each pointer is one member file of a govdocs1 zip (zip_entry gives the local header offset), fetched by byte range"
1297
+ files:
1298
+ - "#00"
1299
+ - "#01"
1300
+ - "#02"
1301
+ - "#03"
1302
+ - "#04"
1303
+ - "#05"
1304
+ - "#06"
1305
+ - "#07"
1306
+ - "#08"
1307
+ - "#09"
1308
+ - "#10"
1309
+ - "#11"
1310
+ - "#12"
1311
+ - "#13"
1312
+ - "#14"
1313
+ - "#15"
1314
+ - "#16"
1315
+ - "#17"
1316
+ - "#18"
1317
+ - "#19"
1318
+ - "#20"
1319
+ - "#21"
1320
+ - "#22"
1321
+ - "#23"
1322
+ - "#24"
1323
+ - "#25"
1324
+ - "#26"
1325
+ - "#27"
1326
+ - "#28"
1327
+ - "#29"
1328
+ sample_n: 30
1329
+ checked_at: "2026-10-04"
1330
+ unit.state: {fill: derived, basis: "packed when pointer_level is archive, else expanded"}
1331
+ unit.usable_unit:
1332
+ fill: judgment
1333
+ basis: "every sample is one standalone file"
1334
+ files:
1335
+ - "#00"
1336
+ - "#01"
1337
+ - "#02"
1338
+ - "#03"
1339
+ - "#04"
1340
+ - "#05"
1341
+ - "#06"
1342
+ - "#07"
1343
+ - "#08"
1344
+ - "#09"
1345
+ - "#10"
1346
+ - "#11"
1347
+ - "#12"
1348
+ - "#13"
1349
+ - "#14"
1350
+ - "#15"
1351
+ - "#16"
1352
+ - "#17"
1353
+ - "#18"
1354
+ - "#19"
1355
+ - "#20"
1356
+ - "#21"
1357
+ - "#22"
1358
+ - "#23"
1359
+ - "#24"
1360
+ - "#25"
1361
+ - "#26"
1362
+ - "#27"
1363
+ - "#28"
1364
+ - "#29"
1365
+ sample_n: 30
1366
+ checked_at: "2026-10-04"
cards/govdocs1.yaml ADDED
@@ -0,0 +1,220 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: govdocs1
8
+ parent: null
9
+ title: "GovDocs1 (Digital Corpora)"
10
+ description: "About 986,000 real files collected from U.S. .gov websites, distributed as 1,000 zip archives of about 1,000 files each; mixed types (pdf, html, jpg, doc, xls, ppt, txt, xml, gif, ps, csv and more). One pointer per zip archive, the smallest unit the source distributes."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: dd16399815605ed1565269ee46abd980f1fdedc4, at: "2026-10-01", rev_candidates: 3}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "Digital Corpora terms"
17
+ licence_url: "https://digitalcorpora.org/corpora/file-corpora/files/"
18
+ licence_statement: "The site states the files are \"freely available for research and may be (to the best of our knowledge) freely redistributed\"."
19
+ per_collection: null
20
+ training_ok: "yes"
21
+ redistribution_ok: "yes"
22
+ attribution: null
23
+ per_file: null
24
+ licence_tiers: {yes/yes: 1000}
25
+ spdx_top: {"(none)": 1000}
26
+ spdx_distinct: 1
27
+ licence_sample_check: null
28
+ unit:
29
+ pointer_level: null
30
+ pointer_count: 1000
31
+ units: {file: 1000}
32
+ state: null
33
+ expanded_count: null
34
+ bytes_total: 333494682740
35
+ rows_without_size: 0
36
+ collections: null
37
+ sub_collections: null
38
+ batches: null
39
+ n_ext: 1
40
+ usable_unit: null
41
+ form:
42
+ by_ext: {zip: 1000}
43
+ bytes_p50: 301001689
44
+ bytes_p95: 583946339
45
+ bytes_max: 1605399030
46
+ size_quantiles:
47
+ max: 1605399030
48
+ min: 148702494
49
+ p25: 254255775
50
+ p50: 301001689
51
+ p75: 368568120
52
+ p90: 481298131
53
+ p95: 583946339
54
+ p99: 797747444
55
+ p99.9: 1417636926
56
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
57
+ by_ext_size:
58
+ - {ext: zip, rows: 1000, bytes: 333494682740, p50: 301001689, p95: 583946339, max: 1605399030}
59
+ over_cap:
60
+ cap: 61944279
61
+ rows_over: 1000
62
+ bytes_over: 333494682740
63
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
64
+ largest:
65
+ - {pointer_id: "govdocs1:zipfiles:657.zip", size_bytes: 1605399030}
66
+ - {pointer_id: "govdocs1:zipfiles:656.zip", size_bytes: 1417636926}
67
+ - {pointer_id: "govdocs1:zipfiles:349.zip", size_bytes: 1165044862}
68
+ - {pointer_id: "govdocs1:zipfiles:125.zip", size_bytes: 1055825900}
69
+ - {pointer_id: "govdocs1:zipfiles:536.zip", size_bytes: 963725474}
70
+ date_years: null
71
+ media_type: {application/zip: 1000}
72
+ facets: null
73
+ artifact: null
74
+ source_origin: null
75
+ parse_cost: null
76
+ parse_quirks: null
77
+ structure:
78
+ self_describing: null
79
+ integrity_builtin: null
80
+ has_reference_output: null
81
+ invariants: null
82
+ tools: null
83
+ ships:
84
+ problem_statement: {present: null}
85
+ reference_solution: {share: null}
86
+ tests: {present: null}
87
+ env_definition: {present: null}
88
+ known_answer: {present: null}
89
+ generator: {present: null}
90
+ upstream_labels: {present: null}
91
+ class: null
92
+ subject:
93
+ domain_span: null
94
+ shape_span: null
95
+ skill_family_span: null
96
+ ont_domain: null
97
+ tech_vocabulary: null
98
+ risk:
99
+ contamination: none
100
+ overlap_check:
101
+ done: true
102
+ pointers_flagged: 0
103
+ pointers_removed: 0
104
+ by_class: {}
105
+ benches: {}
106
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
107
+ pii: null
108
+ pii_evidence: null
109
+ malware: null
110
+ ship_bytes: "yes"
111
+ subset:
112
+ rule: "take all rows; licence, size and version are labels, not filters"
113
+ rows_in_subset: 1000
114
+ last_built:
115
+ date: "2026-10-02"
116
+ rule: "older rule that filtered by licence tier"
117
+ rows: 1000
118
+ rows_in_subset: 1000
119
+ excluded_by_licence: 0
120
+ duplicates: 0
121
+ access:
122
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
123
+ verify: "Compare size_bytes, md5 and sha1 (published). The S3 ETag is a multipart ETag, not an md5."
124
+ hosts: {"s3://digitalcorpora": 1000}
125
+ uri_example: "s3://digitalcorpora/corpora/files/govdocs1/zipfiles/000.zip"
126
+ needs_auth: {"false": 1000}
127
+ credential: null
128
+ hash_source: {source_published: 1000}
129
+ range_supported: null
130
+ reachable:
131
+ index_flag: {"yes": 1000}
132
+ index_check:
133
+ checked_at: "2026-10-01"
134
+ sample: null
135
+ checked: 1000
136
+ ok: 1000
137
+ by_status: {ok: 1000}
138
+ our_sample: null
139
+ drift: null
140
+ sample: null
141
+ evidence:
142
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
143
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
144
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
145
+ class: {fill: unfilled, basis: "ships not judged"}
146
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
147
+ form.by_ext:
148
+ fill: auto
149
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
150
+ checked_at: "2026-10-03"
151
+ form.date_years:
152
+ fill: auto
153
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
154
+ checked_at: "2026-10-03"
155
+ form.over_cap:
156
+ fill: auto
157
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
158
+ checked_at: "2026-10-03"
159
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
160
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
161
+ form.size_quantiles:
162
+ fill: auto
163
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
164
+ checked_at: "2026-10-03"
165
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
166
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
167
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
168
+ provenance.licence_tiers:
169
+ fill: auto
170
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
171
+ checked_at: "2026-10-03"
172
+ provenance.spdx_top:
173
+ fill: auto
174
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
175
+ checked_at: "2026-10-03"
176
+ risk.contamination:
177
+ fill: auto
178
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
179
+ checked_at: "2026-10-03"
180
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
181
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
184
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
186
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
188
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
189
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
190
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
191
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
192
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
193
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
194
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
195
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
196
+ snapshot.rev:
197
+ fill: auto
198
+ basis: "last commit listed in the publish receipt of this source"
199
+ checked_at: "2026-10-03"
200
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ subset:
211
+ fill: auto
212
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
213
+ checked_at: "2026-10-03"
214
+ unit.pointer_count:
215
+ fill: auto
216
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
217
+ checked_at: "2026-10-03"
218
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
220
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/govinfo-bulk-data.yaml ADDED
@@ -0,0 +1,264 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: govinfo-bulk-data
8
+ parent: null
9
+ title: "govinfo bulk data"
10
+ description: "Files from the govinfo Bulk Data Repository of the U.S. Government Publishing Office: bill texts, bill status and summaries, the Federal Register, the Code of Federal Regulations (annual and electronic editions), public laws, the Statutes at Large, presidential documents and more. One pointer per file in the public JSON listings. The source gives sizes but no hashes; compute them at retrieval."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: d90ff1732e4cf02d79a6132a143f5be8bb8b1555, at: "2026-10-01", rev_candidates: 2}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "U.S. Government work (public domain, 17 U.S.C. 105)"
17
+ licence_url: "https://www.govinfo.gov/about/policies#copyright"
18
+ licence_statement: "Works of the U.S. Government are not subject to copyright in the United States (17 U.S.C. 105). The govinfo copyright notice adds that Government publications may contain copyrighted material used with the permission of the copyright owner."
19
+ per_collection: null
20
+ training_ok: "yes"
21
+ redistribution_ok: "yes"
22
+ attribution: null
23
+ per_file: null
24
+ licence_tiers: {yes/yes: 419274}
25
+ spdx_top: {"(none)": 419274}
26
+ spdx_distinct: 1
27
+ licence_sample_check: null
28
+ unit:
29
+ pointer_level: null
30
+ pointer_count: 419274
31
+ units: {file: 419274}
32
+ state: null
33
+ expanded_count: null
34
+ bytes_total: 56366557780
35
+ rows_without_size: 0
36
+ collections:
37
+ distinct: 15
38
+ top:
39
+ BILLS: 138070
40
+ BILLSTATUS: 173182
41
+ BILLSUM: 88627
42
+ CFR: 6596
43
+ COMPS: 2696
44
+ ECFR: 51
45
+ FR: 6677
46
+ PAI: 1018
47
+ PLAW: 2167
48
+ STATUTE: 139
49
+ sub_collections: null
50
+ batches: null
51
+ n_ext: 12
52
+ usable_unit: null
53
+ form:
54
+ by_ext:
55
+ xml: 419181
56
+ zip: 27
57
+ html: 18
58
+ jpg: 17
59
+ xsl: 8
60
+ pdf: 6
61
+ dtd: 4
62
+ xsd: 4
63
+ gif: 3
64
+ htm: 3
65
+ css: 2
66
+ txt: 1
67
+ bytes_p50: 7992
68
+ bytes_p95: 104865
69
+ bytes_max: 161213870
70
+ size_quantiles:
71
+ max: 161213870
72
+ min: 303
73
+ p25: 3908
74
+ p50: 7992
75
+ p75: 14744
76
+ p90: 35007
77
+ p95: 104865
78
+ p99: 3939432
79
+ p99.9: 7976556
80
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
81
+ by_ext_size:
82
+ - {ext: xml, rows: 419181, bytes: 55951014348, p50: 7991, p95: 104369, max: 161213870}
83
+ - {ext: zip, rows: 27, bytes: 411877289, p50: 3540225, p95: 60093265, max: 60674886}
84
+ - {ext: html, rows: 18, bytes: 17399, p50: 1093, p95: 1096, max: 1096}
85
+ - {ext: jpg, rows: 17, bytes: 208383, p50: 12188, p95: 21149, max: 21149}
86
+ - {ext: xsl, rows: 8, bytes: 1234018, p50: 18394, p95: 949726, max: 949726}
87
+ - {ext: pdf, rows: 6, bytes: 970704, p50: 134532, p95: 254318, max: 254318}
88
+ - {ext: dtd, rows: 4, bytes: 528571, p50: 62249, p95: 270543, max: 270543}
89
+ - {ext: xsd, rows: 4, bytes: 626615, p50: 103290, p95: 258783, max: 258783}
90
+ - {ext: gif, rows: 3, bytes: 24097, p50: 806, p95: 22988, max: 22988}
91
+ - {ext: htm, rows: 3, bytes: 3558, p50: 1186, p95: 1186, max: 1186}
92
+ - {ext: css, rows: 2, bytes: 52048, p50: 25318, p95: 26730, max: 26730}
93
+ - {ext: txt, rows: 1, bytes: 750, p50: 750, p95: 750, max: 750}
94
+ over_cap:
95
+ cap: 61944279
96
+ rows_over: 3
97
+ bytes_over: 320877408
98
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
99
+ largest:
100
+ - {pointer_id: "govinfo-bulk-data:ECFR:title-40/ECFR-title40.xml", size_bytes: 161213870}
101
+ - {pointer_id: "govinfo-bulk-data:ECFR:title-26/ECFR-title26.xml", size_bytes: 88045995}
102
+ - {pointer_id: "govinfo-bulk-data:STATUTE:44/STATUTE-44.xml", size_bytes: 71617543}
103
+ - {pointer_id: "govinfo-bulk-data:CBD:1999/CBD-1999.zip", size_bytes: 60674886}
104
+ - {pointer_id: "govinfo-bulk-data:CBD:1998/CBD-1998.zip", size_bytes: 60093265}
105
+ date_years: null
106
+ media_type:
107
+ "(none)": 16
108
+ application/pdf: 6
109
+ application/xml: 419181
110
+ application/zip: 27
111
+ image/gif: 3
112
+ image/jpeg: 17
113
+ text/css: 2
114
+ text/html: 21
115
+ text/plain: 1
116
+ facets: null
117
+ artifact: null
118
+ source_origin: null
119
+ parse_cost: null
120
+ parse_quirks: null
121
+ structure:
122
+ self_describing: null
123
+ integrity_builtin: null
124
+ has_reference_output: null
125
+ invariants: null
126
+ tools: null
127
+ ships:
128
+ problem_statement: {present: null}
129
+ reference_solution: {share: null}
130
+ tests: {present: null}
131
+ env_definition: {present: null}
132
+ known_answer: {present: null}
133
+ generator: {present: null}
134
+ upstream_labels: {present: null}
135
+ class: null
136
+ subject:
137
+ domain_span: null
138
+ shape_span: null
139
+ skill_family_span: null
140
+ ont_domain: null
141
+ tech_vocabulary: null
142
+ risk:
143
+ contamination: none
144
+ overlap_check:
145
+ done: true
146
+ pointers_flagged: 0
147
+ pointers_removed: 0
148
+ by_class: {}
149
+ benches: {}
150
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
151
+ pii: null
152
+ pii_evidence: null
153
+ malware: null
154
+ ship_bytes: "yes"
155
+ subset:
156
+ rule: "take all rows; licence, size and version are labels, not filters"
157
+ rows_in_subset: 419274
158
+ last_built:
159
+ date: "2026-10-02"
160
+ rule: "older rule that filtered by licence tier"
161
+ rows: 419274
162
+ rows_in_subset: 419274
163
+ excluded_by_licence: 0
164
+ duplicates: 0
165
+ access:
166
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
167
+ verify: "Compare size_bytes (from the listing). No hash is published (empty hashes, computed_on_fetch): a download of the right size counts as ok; record the measured sha256. The source regenerates some files (bill status is updated daily); on a size mismatch, compare source_meta.last_modified."
168
+ hosts: {"https://www.govinfo.gov": 419274}
169
+ uri_example: "https://www.govinfo.gov/bulkdata/BILLS/113/1/hconres/BILLS-113hconres10ih.xml"
170
+ needs_auth: {"false": 419274}
171
+ credential: null
172
+ hash_source: {computed_on_fetch: 419274}
173
+ range_supported: null
174
+ reachable:
175
+ index_flag: {unknown: 419274}
176
+ index_check:
177
+ checked_at: "2026-10-01"
178
+ sample: 30
179
+ checked: 351
180
+ ok: 31
181
+ by_status: {error: 320, ok: 31}
182
+ our_sample: null
183
+ drift: null
184
+ sample: null
185
+ evidence:
186
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
188
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
189
+ class: {fill: unfilled, basis: "ships not judged"}
190
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
191
+ form.by_ext:
192
+ fill: auto
193
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
194
+ checked_at: "2026-10-03"
195
+ form.date_years:
196
+ fill: auto
197
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
198
+ checked_at: "2026-10-03"
199
+ form.over_cap:
200
+ fill: auto
201
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
202
+ checked_at: "2026-10-03"
203
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ form.size_quantiles:
206
+ fill: auto
207
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
208
+ checked_at: "2026-10-03"
209
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
211
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
212
+ provenance.licence_tiers:
213
+ fill: auto
214
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
215
+ checked_at: "2026-10-03"
216
+ provenance.spdx_top:
217
+ fill: auto
218
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
219
+ checked_at: "2026-10-03"
220
+ risk.contamination:
221
+ fill: auto
222
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
223
+ checked_at: "2026-10-03"
224
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
228
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
237
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
238
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
239
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
240
+ snapshot.rev:
241
+ fill: auto
242
+ basis: "last commit listed in the publish receipt of this source"
243
+ checked_at: "2026-10-03"
244
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
245
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
247
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
249
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
250
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
251
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
252
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
253
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ subset:
255
+ fill: auto
256
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
257
+ checked_at: "2026-10-03"
258
+ unit.pointer_count:
259
+ fill: auto
260
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
261
+ checked_at: "2026-10-03"
262
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
264
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/hackage-packages.yaml ADDED
@@ -0,0 +1,239 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: hackage-packages
8
+ parent: null
9
+ title: "Hackage packages"
10
+ description: "Source packages (.tar.gz) from Hackage, the central package repository for Haskell, one pointer per package version. Versions are read from the package index 01-index.tar.gz (Last-Modified Fri, 02 Oct 2026 22:18:55 GMT). SHA-256, MD5 and sizes come from the package.json entry of each version in the index; licenses and other metadata come from the latest revision of the version's .cabal file. Packages can contain malicious code; treat files as data and run nothing outside a sandbox."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-02"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per package (license field of the .cabal file)"
20
+ licence_url: "https://hackage.haskell.org/upload"
21
+ licence_statement: "Hackage gives no license of its own; each package declares its license. Its upload policy says material uploaded and distributed via the site must be under an open source license."
22
+ per_collection: null
23
+ training_ok: conditional
24
+ redistribution_ok: conditional
25
+ attribution: null
26
+ per_file: "license_override.notes gives the license field of the .cabal file; license_override.spdx gives the SPDX id where one is recognised. Old style names are mapped as Cabal maps them (GPL-3 to GPL-3.0-only, BSD3 to BSD-3-Clause, and so on). AllRightsReserved is marked not for training; OtherLicense, a missing field, GPL or LGPL without a version and unrecognised values are marked unknown."
27
+ licence_tiers: {unknown/unknown: 11518, yes/conditional: 13146, yes/yes: 129463}
28
+ spdx_top:
29
+ "(none)": 11518
30
+ "0BSD": 93
31
+ AGPL-3.0-only: 829
32
+ AGPL-3.0-or-later: 252
33
+ Apache-2.0: 4781
34
+ "Apache-2.0 OR BSD-3-Clause": 15
35
+ "Apache-2.0 OR MPL-2.0": 42
36
+ BSD-2-Clause: 1536
37
+ BSD-2-Clause-Patent: 422
38
+ BSD-3-Clause: 86905
39
+ "BSD-3-Clause OR Apache-2.0": 16
40
+ CC0-1.0: 100
41
+ EUPL-1.2: 38
42
+ GPL-2.0-only: 1562
43
+ GPL-2.0-or-later: 322
44
+ spdx_distinct: 53
45
+ licence_sample_check: null
46
+ unit:
47
+ pointer_level: null
48
+ pointer_count: 154127
49
+ units: {file: 154127}
50
+ state: null
51
+ expanded_count: null
52
+ bytes_total: 20003722578
53
+ rows_without_size: 0
54
+ collections:
55
+ distinct: 19515
56
+ top:
57
+ egison: 194
58
+ git-annex: 317
59
+ hakyll: 202
60
+ haskoin-store: 321
61
+ hlint: 223
62
+ pandoc: 224
63
+ persistent: 197
64
+ purescript: 205
65
+ warp: 222
66
+ yesod-core: 224
67
+ sub_collections: null
68
+ batches: null
69
+ n_ext: 1
70
+ usable_unit: null
71
+ form:
72
+ by_ext: {gz: 154127}
73
+ bytes_p50: 14724
74
+ bytes_p95: 305819
75
+ bytes_max: 46198466
76
+ size_quantiles:
77
+ max: 46198466
78
+ min: 295
79
+ p25: 6156
80
+ p50: 14724
81
+ p75: 39093
82
+ p90: 115023
83
+ p95: 305819
84
+ p99: 2317914
85
+ p99.9: 12897169
86
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
87
+ by_ext_size:
88
+ - {ext: gz, rows: 154127, bytes: 20003722578, p50: 14724, p95: 305819, max: 46198466}
89
+ over_cap:
90
+ cap: 61944279
91
+ rows_over: 0
92
+ bytes_over: 0
93
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
94
+ largest:
95
+ - {pointer_id: "hackage-packages:gf:gf-3.9.tar.gz", size_bytes: 46198466}
96
+ - {pointer_id: "hackage-packages:gf:gf-3.8.tar.gz", size_bytes: 45893872}
97
+ - {pointer_id: "hackage-packages:gf:gf-3.7.1.tar.gz", size_bytes: 45739522}
98
+ - {pointer_id: "hackage-packages:hOpenPGP:hOpenPGP-2.5.1.tar.gz", size_bytes: 44700255}
99
+ - {pointer_id: "hackage-packages:hOpenPGP:hOpenPGP-2.5.tar.gz", size_bytes: 44697429}
100
+ date_years: null
101
+ media_type: {"(none)": 154127}
102
+ facets: null
103
+ artifact: null
104
+ source_origin: null
105
+ parse_cost: null
106
+ parse_quirks: null
107
+ structure:
108
+ self_describing: null
109
+ integrity_builtin: null
110
+ has_reference_output: null
111
+ invariants: null
112
+ tools: null
113
+ ships:
114
+ problem_statement: {present: null}
115
+ reference_solution: {share: null}
116
+ tests: {present: null}
117
+ env_definition: {present: null}
118
+ known_answer: {present: null}
119
+ generator: {present: null}
120
+ upstream_labels: {present: null}
121
+ class: null
122
+ subject:
123
+ domain_span: null
124
+ shape_span: null
125
+ skill_family_span: null
126
+ ont_domain: null
127
+ tech_vocabulary: null
128
+ risk:
129
+ contamination: none
130
+ overlap_check:
131
+ done: true
132
+ pointers_flagged: 0
133
+ pointers_removed: 0
134
+ by_class: {}
135
+ benches: {}
136
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
137
+ pii: null
138
+ pii_evidence: null
139
+ malware: null
140
+ ship_bytes: conditional
141
+ subset:
142
+ rule: "take all rows; licence, size and version are labels, not filters"
143
+ rows_in_subset: 154127
144
+ last_built: null
145
+ access:
146
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
147
+ verify: "Compare size_bytes and sha256 (published in the index); md5 is also published."
148
+ hosts: {"https://hackage.haskell.org": 154127}
149
+ uri_example: "https://hackage.haskell.org/package/2captcha-0.1.0.0/2captcha-0.1.0.0.tar.gz"
150
+ needs_auth: {"false": 154127}
151
+ credential: null
152
+ hash_source: {source_published: 154127}
153
+ range_supported: null
154
+ reachable:
155
+ index_flag: {unknown: 154127}
156
+ index_check: null
157
+ our_sample: null
158
+ drift: null
159
+ sample: null
160
+ evidence:
161
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
162
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
163
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
164
+ class: {fill: unfilled, basis: "ships not judged"}
165
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
166
+ form.by_ext:
167
+ fill: auto
168
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
169
+ checked_at: "2026-10-03"
170
+ form.date_years:
171
+ fill: auto
172
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
173
+ checked_at: "2026-10-03"
174
+ form.over_cap:
175
+ fill: auto
176
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
177
+ checked_at: "2026-10-03"
178
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
179
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
180
+ form.size_quantiles:
181
+ fill: auto
182
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
183
+ checked_at: "2026-10-03"
184
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
186
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ provenance.licence_tiers:
188
+ fill: auto
189
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
190
+ checked_at: "2026-10-03"
191
+ provenance.spdx_top:
192
+ fill: auto
193
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
194
+ checked_at: "2026-10-03"
195
+ risk.contamination:
196
+ fill: auto
197
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
198
+ checked_at: "2026-10-03"
199
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
203
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
212
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
213
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
214
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ snapshot.rev:
216
+ fill: auto
217
+ basis: "last commit listed in the publish receipt of this source"
218
+ checked_at: "2026-10-03"
219
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
228
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ subset:
230
+ fill: auto
231
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
232
+ checked_at: "2026-10-03"
233
+ unit.pointer_count:
234
+ fill: auto
235
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
236
+ checked_at: "2026-10-03"
237
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
238
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
239
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/hacker-news-threads.yaml ADDED
@@ -0,0 +1,234 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: hacker-news-threads
8
+ parent: null
9
+ title: "Hacker News discussion threads"
10
+ description: "Stories posted to Hacker News (news.ycombinator.com), including Ask HN and Show HN posts, listed newest first by creation time through the HN Search API run by Algolia (search_by_date with the story tag, in time windows of at most 1000 stories). Stories created from 2006-10-01 00:00 UTC up to 2026-10-03 00:00 UTC were in scope; the oldest story listed was created at 2006-10-09 18:21 UTC. One pointer per story; the uri returns the story together with its whole comment tree as JSON. Title, author, points, comment count and topic tags matched in the title are in source_meta. Collections are the UTC year the story was created."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "No content license; Y Combinator Terms of Use"
20
+ licence_url: "https://www.ycombinator.com/legal/"
21
+ licence_statement: "Stories and comments belong to their authors, who grant a license to Y Combinator only; no license is granted to other users. The Y Combinator Terms of Use, which cover all of its subdomains including news.ycombinator.com, restrict copying, scraping, distribution, derivative works and data mining of site content (quoted in notes). They contain no clause about machine learning training."
22
+ per_collection: null
23
+ training_ok: "no"
24
+ redistribution_ok: "no"
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {no/no: 4790481}
28
+ spdx_top: {"(none)": 4790481}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 4790481
34
+ units: {file: 4790481}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 0
38
+ rows_without_size: 4790481
39
+ collections:
40
+ distinct: 21
41
+ top:
42
+ "2016": 283371
43
+ "2017": 286683
44
+ "2019": 300866
45
+ "2020": 339459
46
+ "2021": 290676
47
+ "2022": 299091
48
+ "2023": 317353
49
+ "2024": 297589
50
+ "2025": 301530
51
+ "2026": 279716
52
+ sub_collections: null
53
+ batches: null
54
+ n_ext: 1
55
+ usable_unit: null
56
+ form:
57
+ by_ext: {json: 4790481}
58
+ bytes_p50: null
59
+ bytes_p95: null
60
+ bytes_max: null
61
+ size_quantiles: null
62
+ by_ext_size:
63
+ - {ext: json, rows: 4790481, bytes: 0, p50: null, p95: null, max: null}
64
+ over_cap:
65
+ cap: 61944279
66
+ rows_over: 0
67
+ bytes_over: 0
68
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
69
+ largest: []
70
+ date_years:
71
+ key: created_at
72
+ rows_with_date: 4790481
73
+ counts:
74
+ "2006": 46
75
+ "2007": 21663
76
+ "2008": 61554
77
+ "2009": 88327
78
+ "2010": 140678
79
+ "2011": 199055
80
+ "2012": 232835
81
+ "2013": 260583
82
+ "2014": 247554
83
+ "2015": 265007
84
+ "2016": 283371
85
+ "2017": 286683
86
+ "2018": 276845
87
+ "2019": 300866
88
+ "2020": 339459
89
+ "2021": 290676
90
+ "2022": 299091
91
+ "2023": 317353
92
+ "2024": 297589
93
+ "2025": 301530
94
+ "2026": 279716
95
+ media_type: {application/json: 4790481}
96
+ facets:
97
+ type: {ask_hn: 280626, show_hn: 210333, story: 4299522}
98
+ artifact: null
99
+ source_origin: null
100
+ parse_cost: null
101
+ parse_quirks: null
102
+ structure:
103
+ self_describing: null
104
+ integrity_builtin: null
105
+ has_reference_output: null
106
+ invariants: null
107
+ tools: null
108
+ ships:
109
+ problem_statement: {present: null}
110
+ reference_solution: {share: null}
111
+ tests: {present: null}
112
+ env_definition: {present: null}
113
+ known_answer: {present: null}
114
+ generator: {present: null}
115
+ upstream_labels: {present: null}
116
+ class: null
117
+ subject:
118
+ domain_span: null
119
+ shape_span: null
120
+ skill_family_span: null
121
+ ont_domain: null
122
+ tech_vocabulary: null
123
+ risk:
124
+ contamination: none
125
+ overlap_check:
126
+ done: true
127
+ pointers_flagged: 0
128
+ pointers_removed: 0
129
+ by_class: {}
130
+ benches: {}
131
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
132
+ pii: null
133
+ pii_evidence: null
134
+ malware: null
135
+ ship_bytes: "no"
136
+ subset:
137
+ rule: "take all rows; licence, size and version are labels, not filters"
138
+ rows_in_subset: 4790481
139
+ last_built: null
140
+ access:
141
+ method: "GET uri; the response is JSON (the story and all of its comments, nested). The HN Search API allows 10,000 requests per hour from one IP address."
142
+ verify: "No size or hash is published, and a thread changes as comments are added, edited or deleted. Record the size and sha256 of the downloaded bytes (hash_source computed_on_fetch) and check that they parse as JSON."
143
+ hosts: {"https://hn.algolia.com": 4790481}
144
+ uri_example: "https://hn.algolia.com/api/v1/items/1"
145
+ needs_auth: {"false": 4790481}
146
+ credential: null
147
+ hash_source: {computed_on_fetch: 4790481}
148
+ range_supported: null
149
+ reachable:
150
+ index_flag: {unknown: 4790481}
151
+ index_check: null
152
+ our_sample: null
153
+ drift: null
154
+ sample: null
155
+ evidence:
156
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
157
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
158
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
159
+ class: {fill: unfilled, basis: "ships not judged"}
160
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
161
+ form.by_ext:
162
+ fill: auto
163
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
164
+ checked_at: "2026-10-03"
165
+ form.date_years:
166
+ fill: auto
167
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
168
+ checked_at: "2026-10-03"
169
+ form.over_cap:
170
+ fill: auto
171
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
172
+ checked_at: "2026-10-03"
173
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
174
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
175
+ form.size_quantiles:
176
+ fill: auto
177
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
178
+ checked_at: "2026-10-03"
179
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
180
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
181
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ provenance.licence_tiers:
183
+ fill: auto
184
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
185
+ checked_at: "2026-10-03"
186
+ provenance.spdx_top:
187
+ fill: auto
188
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
189
+ checked_at: "2026-10-03"
190
+ risk.contamination:
191
+ fill: auto
192
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
193
+ checked_at: "2026-10-03"
194
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
195
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
196
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
197
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
198
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
199
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
202
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
206
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ snapshot.rev:
211
+ fill: auto
212
+ basis: "last commit listed in the publish receipt of this source"
213
+ checked_at: "2026-10-03"
214
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
222
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ subset:
225
+ fill: auto
226
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
227
+ checked_at: "2026-10-03"
228
+ unit.pointer_count:
229
+ fill: auto
230
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
231
+ checked_at: "2026-10-03"
232
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
234
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/harvard-dataverse.yaml ADDED
@@ -0,0 +1,293 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: harvard-dataverse
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: null
20
+ licence_url: null
21
+ licence_statement: null
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers:
28
+ conditional/conditional: 133219
29
+ conditional/no: 70069
30
+ unknown/no: 370
31
+ yes/conditional: 21750
32
+ yes/yes: 129392
33
+ spdx_top:
34
+ CC-BY-4.0: 129392
35
+ CC-BY-NC-4.0: 66706
36
+ CC-BY-NC-ND-4.0: 70069
37
+ CC-BY-NC-SA-4.0: 66513
38
+ CC-BY-ND-4.0: 370
39
+ CC-BY-SA-4.0: 21750
40
+ spdx_distinct: 6
41
+ licence_sample_check: null
42
+ unit:
43
+ pointer_level: null
44
+ pointer_count: 354800
45
+ units: {file: 354800}
46
+ state: null
47
+ expanded_count: null
48
+ bytes_total: 35654098705673
49
+ rows_without_size: 0
50
+ collections:
51
+ distinct: 3847
52
+ top:
53
+ "10.7910/DVN/OJXIS1": 12267
54
+ "10.7910/DVN/OQCPWF": 14076
55
+ "10.7910/DVN/OYQIUI": 5571
56
+ "10.7910/DVN/ROFOT5": 14270
57
+ "10.7910/DVN/SOBQ4R": 14451
58
+ "10.7910/DVN/UAABZK": 11159
59
+ "10.7910/DVN/V2C6G2": 10523
60
+ "10.7910/DVN/VDHNVA": 5500
61
+ "10.7910/DVN/YHXJSU": 6811
62
+ "10.7910/DVN/Z0VXLI": 11297
63
+ sub_collections: null
64
+ batches: null
65
+ n_ext: 2483
66
+ usable_unit: null
67
+ form:
68
+ by_ext:
69
+ jpg: 70195
70
+ pdf: 32428
71
+ stl: 28498
72
+ mat: 20012
73
+ nc: 17328
74
+ zip: 16319
75
+ tif: 15358
76
+ txt: 14110
77
+ csv: 13565
78
+ pt: 11347
79
+ mrc: 10946
80
+ gz: 6818
81
+ wav: 6537
82
+ parquet: 5592
83
+ mp4: 5165
84
+ png: 4951
85
+ pdb: 4933
86
+ h5: 4301
87
+ m4a: 3946
88
+ xlsx: 3674
89
+ "(2463 other types)": 58777
90
+ bytes_p50: 2143572
91
+ bytes_p95: 437770668
92
+ bytes_max: 184838663956
93
+ size_quantiles:
94
+ max: 184838663956
95
+ min: 0
96
+ p25: 133535
97
+ p50: 2143572
98
+ p75: 10804284
99
+ p90: 108001024
100
+ p95: 437770668
101
+ p99: 1972249364
102
+ p99.9: 9746329210
103
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
104
+ by_ext_size:
105
+ - {ext: jpg, rows: 70195, bytes: 146148087049, p50: 528999, p95: 8434353, max: 181122782}
106
+ - {ext: pdf, rows: 32428, bytes: 121044638605, p50: 204939, p95: 8405926, max: 898775487}
107
+ - {ext: stl, rows: 28498, bytes: 222607681632, p50: 6588884, p95: 15098084, max: 24474484}
108
+ - {ext: mat, rows: 20012, bytes: 420146766692, p50: 12294570, p95: 80156691, max: 2328714772}
109
+ - {ext: nc, rows: 17328, bytes: 4728097653380, p50: 3384436, p95: 1885620451, max: 1885620511}
110
+ - {ext: zip, rows: 16319, bytes: 7612603312311, p50: 24039844, p95: 1977642487, max: 168741970393}
111
+ - {ext: tif, rows: 15358, bytes: 1554936428535, p50: 1407854, p95: 679986098, max: 2634563738}
112
+ - {ext: txt, rows: 14110, bytes: 10849039763, p50: 2733, p95: 67741, max: 644794355}
113
+ - {ext: csv, rows: 13565, bytes: 443178523443, p50: 170477, p95: 110024620, max: 12196007027}
114
+ - {ext: pt, rows: 11347, bytes: 404179759386, p50: 10853076, p95: 144964807, max: 2680882511}
115
+ - {ext: mrc, rows: 10946, bytes: 1399395940380, p50: 95552512, p95: 340737024, max: 2396309452}
116
+ - {ext: gz, rows: 6818, bytes: 3847674250863, p50: 8954529, p95: 2401944576, max: 32631939404}
117
+ - {ext: wav, rows: 6537, bytes: 35477815647, p50: 4149440, p95: 6816992, max: 345600044}
118
+ - {ext: parquet, rows: 5592, bytes: 316026681468, p50: 15374543, p95: 62151310, max: 19733203305}
119
+ - {ext: mp4, rows: 5165, bytes: 337859464980, p50: 50879774, p95: 128059508, max: 3438503413}
120
+ - {ext: png, rows: 4951, bytes: 10919393068, p50: 237763, p95: 20207143, max: 80221227}
121
+ - {ext: pdb, rows: 4933, bytes: 5988907382, p50: 850014, p95: 3637467, max: 29037933}
122
+ - {ext: h5, rows: 4301, bytes: 218946657266, p50: 33562580, p95: 40072228, max: 2308770048}
123
+ - {ext: m4a, rows: 3946, bytes: 2630532273, p50: 257083, p95: 2589727, max: 95040990}
124
+ - {ext: xlsx, rows: 3674, bytes: 13477264995, p50: 60151, p95: 18150971, max: 928806211}
125
+ over_cap:
126
+ cap: 61944279
127
+ rows_over: 42561
128
+ bytes_over: 33968021882174
129
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
130
+ largest:
131
+ - {pointer_id: "harvard-dataverse:10.7910/DVN/XMRBWN:10551815/data.tar.xz", size_bytes: 184838663956}
132
+ - {pointer_id: "harvard-dataverse:10.7910/DVN/FCDG0W:7522504/EMD_2.zip", size_bytes: 168741970393}
133
+ - {pointer_id: "harvard-dataverse:10.7910/DVN/FCDG0W:7506250/EMD_3.zip", size_bytes: 94056286344}
134
+ - pointer_id: "harvard-dataverse:10.7910/DVN/CGUENL:7506168/small_dataset.zip"
135
+ size_bytes: 93801532820
136
+ - pointer_id: "harvard-dataverse:10.7910/DVN/AWDKXC:11122644/Charts-1927-1937.tar.xz"
137
+ size_bytes: 73673975544
138
+ date_years: null
139
+ media_type:
140
+ application/marc: 10946
141
+ application/matlab-mat: 19870
142
+ application/netcdf: 13165
143
+ application/octet-stream: 47319
144
+ application/pdf: 32457
145
+ application/vnd.ms-pki.stl: 28496
146
+ application/zip: 15498
147
+ image/jpeg: 72626
148
+ image/tiff: 17047
149
+ text/plain: 15066
150
+ facets: null
151
+ artifact: null
152
+ source_origin: null
153
+ parse_cost: null
154
+ parse_quirks: null
155
+ structure:
156
+ self_describing: null
157
+ integrity_builtin: null
158
+ has_reference_output: null
159
+ invariants: null
160
+ tools: null
161
+ ships:
162
+ problem_statement: {present: null}
163
+ reference_solution: {share: null}
164
+ tests: {present: null}
165
+ env_definition: {present: null}
166
+ known_answer: {present: null}
167
+ generator: {present: null}
168
+ upstream_labels: {present: null}
169
+ class: null
170
+ subject:
171
+ domain_span: null
172
+ shape_span: null
173
+ skill_family_span: null
174
+ ont_domain: null
175
+ tech_vocabulary: null
176
+ risk:
177
+ contamination: none
178
+ overlap_check:
179
+ done: true
180
+ pointers_flagged: 5
181
+ pointers_removed: 0
182
+ by_class: {generic_text: 5}
183
+ benches: {"FrontierSWE v2": 5}
184
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
185
+ pii: null
186
+ pii_evidence: null
187
+ malware: null
188
+ ship_bytes: unknown
189
+ subset:
190
+ rule: "take all rows; licence, size and version are labels, not filters"
191
+ rows_in_subset: 354800
192
+ last_built:
193
+ date: "2026-10-02"
194
+ rule: "older rule that filtered by licence tier"
195
+ rows: 354800
196
+ rows_in_subset: 343202
197
+ excluded_by_licence: 370
198
+ duplicates: 11285
199
+ access:
200
+ method: null
201
+ verify: null
202
+ hosts: {"https://dataverse.harvard.edu": 354800}
203
+ uri_example: "https://dataverse.harvard.edu/api/access/datafile/11140712?format=original"
204
+ needs_auth: {"false": 320238, "true": 34562}
205
+ credential: null
206
+ hash_source: {source_published: 354800}
207
+ range_supported: null
208
+ reachable:
209
+ index_flag: {"(none)": 354800}
210
+ index_check: null
211
+ our_sample: null
212
+ drift: null
213
+ sample: null
214
+ evidence:
215
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ class: {fill: unfilled, basis: "ships not judged"}
219
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ form.by_ext:
221
+ fill: auto
222
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
223
+ checked_at: "2026-10-03"
224
+ form.date_years:
225
+ fill: auto
226
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
227
+ checked_at: "2026-10-03"
228
+ form.over_cap:
229
+ fill: auto
230
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
231
+ checked_at: "2026-10-03"
232
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
234
+ form.size_quantiles:
235
+ fill: auto
236
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
237
+ checked_at: "2026-10-03"
238
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
239
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
240
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
241
+ provenance.licence_tiers:
242
+ fill: auto
243
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
244
+ checked_at: "2026-10-03"
245
+ provenance.spdx_top:
246
+ fill: auto
247
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
248
+ checked_at: "2026-10-03"
249
+ risk.contamination:
250
+ fill: auto
251
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
252
+ checked_at: "2026-10-03"
253
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
257
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
260
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
261
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
262
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
265
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
266
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
268
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ snapshot.rev:
270
+ fill: auto
271
+ basis: "last commit listed in the publish receipt of this source"
272
+ checked_at: "2026-10-03"
273
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
275
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ subset:
284
+ fill: auto
285
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
286
+ checked_at: "2026-10-03"
287
+ unit.pointer_count:
288
+ fill: auto
289
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
290
+ checked_at: "2026-10-03"
291
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
293
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/hyperkitty-mailing-list-archives.yaml ADDED
@@ -0,0 +1,211 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: hyperkitty-mailing-list-archives
8
+ parent: null
9
+ title: "Mailman 3 HyperKitty mailing list archives"
10
+ description: "Monthly mbox.gz exports of all public mailing lists on 2 Mailman 3 HyperKitty archives (mail.python.org, lists.fedoraproject.org). One pointer per list per month."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "No content license stated"
20
+ licence_url: "https://docs.mailman3.org/projects/hyperkitty/"
21
+ licence_statement: "Messages remain the copyright of their senders; the archives state no content license, and no archive terms of use were found on either site."
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {unknown/unknown: 80365}
28
+ spdx_top: {"(none)": 80365}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 80365
34
+ units: {file: 80365}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 0
38
+ rows_without_size: 80365
39
+ collections:
40
+ distinct: 761
41
+ top:
42
+ "astropy@python.org": 329
43
+ "db-sig@python.org": 322
44
+ "edu-sig@python.org": 320
45
+ "mailman-developers@python.org": 353
46
+ "mailman-i18n@python.org": 320
47
+ "mailman-users@python.org": 333
48
+ "numpy-discussion@python.org": 321
49
+ "python-announce-list@python.org": 330
50
+ "python-list@python.org": 332
51
+ "tutor@python.org": 326
52
+ sub_collections: null
53
+ batches: null
54
+ n_ext: 1
55
+ usable_unit: null
56
+ form:
57
+ by_ext: {gz: 80365}
58
+ bytes_p50: null
59
+ bytes_p95: null
60
+ bytes_max: null
61
+ size_quantiles: null
62
+ by_ext_size:
63
+ - {ext: gz, rows: 80365, bytes: 0, p50: null, p95: null, max: null}
64
+ over_cap:
65
+ cap: 61944279
66
+ rows_over: 0
67
+ bytes_over: 0
68
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
69
+ largest: []
70
+ date_years: null
71
+ media_type: {application/gzip: 80365}
72
+ facets:
73
+ domain: {"false": 49897, "true": 30468}
74
+ instance: {lists.fedoraproject.org: 57483, mail.python.org: 22882}
75
+ artifact: null
76
+ source_origin: null
77
+ parse_cost: null
78
+ parse_quirks: null
79
+ structure:
80
+ self_describing: null
81
+ integrity_builtin: null
82
+ has_reference_output: null
83
+ invariants: null
84
+ tools: null
85
+ ships:
86
+ problem_statement: {present: null}
87
+ reference_solution: {share: null}
88
+ tests: {present: null}
89
+ env_definition: {present: null}
90
+ known_answer: {present: null}
91
+ generator: {present: null}
92
+ upstream_labels: {present: null}
93
+ class: null
94
+ subject:
95
+ domain_span: null
96
+ shape_span: null
97
+ skill_family_span: null
98
+ ont_domain: null
99
+ tech_vocabulary: null
100
+ risk:
101
+ contamination: none
102
+ overlap_check:
103
+ done: true
104
+ pointers_flagged: 0
105
+ pointers_removed: 0
106
+ by_class: {}
107
+ benches: {}
108
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
109
+ pii: null
110
+ pii_evidence: null
111
+ malware: null
112
+ ship_bytes: unknown
113
+ subset:
114
+ rule: "take all rows; licence, size and version are labels, not filters"
115
+ rows_in_subset: 80365
116
+ last_built: null
117
+ access:
118
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
119
+ verify: "Download the month's mbox.gz export; record its size and sha256 and check that it decompresses as gzip. The source publishes no size or hash and generates the export on request."
120
+ hosts: {"https://lists.fedoraproject.org": 57483, "https://mail.python.org": 22882}
121
+ uri_example: "https://lists.fedoraproject.org/archives/list/389-announce@lists.fedoraproject.org/export/389-announce@lists.fedoraproject.org-2005-06.mbox.gz?start=2005-06-01&end=2005-07-01"
122
+ needs_auth: {"false": 80365}
123
+ credential: null
124
+ hash_source: {computed_on_fetch: 80365}
125
+ range_supported: null
126
+ reachable:
127
+ index_flag: {unknown: 80365}
128
+ index_check: null
129
+ our_sample: null
130
+ drift: null
131
+ sample: null
132
+ evidence:
133
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
134
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
135
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
136
+ class: {fill: unfilled, basis: "ships not judged"}
137
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
138
+ form.by_ext:
139
+ fill: auto
140
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
141
+ checked_at: "2026-10-03"
142
+ form.date_years:
143
+ fill: auto
144
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
145
+ checked_at: "2026-10-03"
146
+ form.over_cap:
147
+ fill: auto
148
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
149
+ checked_at: "2026-10-03"
150
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
151
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
152
+ form.size_quantiles:
153
+ fill: auto
154
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
155
+ checked_at: "2026-10-03"
156
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
157
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
158
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
159
+ provenance.licence_tiers:
160
+ fill: auto
161
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
162
+ checked_at: "2026-10-03"
163
+ provenance.spdx_top:
164
+ fill: auto
165
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
166
+ checked_at: "2026-10-03"
167
+ risk.contamination:
168
+ fill: auto
169
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
170
+ checked_at: "2026-10-03"
171
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
172
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
173
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
174
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
175
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
176
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
177
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
178
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
179
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
180
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
181
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
184
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
185
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
186
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
187
+ snapshot.rev:
188
+ fill: auto
189
+ basis: "last commit listed in the publish receipt of this source"
190
+ checked_at: "2026-10-03"
191
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
192
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
193
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
194
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
195
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
196
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
197
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
198
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
199
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
200
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
201
+ subset:
202
+ fill: auto
203
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
204
+ checked_at: "2026-10-03"
205
+ unit.pointer_count:
206
+ fill: auto
207
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
208
+ checked_at: "2026-10-03"
209
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
211
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/if-archive.yaml ADDED
@@ -0,0 +1,292 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: if-archive
8
+ parent: null
9
+ title: "Interactive Fiction Archive"
10
+ description: "Files in the Interactive Fiction Archive (ifarchive.org): text adventure games, walkthroughs and hints, interpreters, authoring tools, and related documents and newsgroup archives. One pointer per archive file, listed from the archive's master index (Master-Index.xml). Collections are the archive's top-level directories."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per file (set by each author)"
20
+ licence_url: "https://ifarchive.org/misc/license.html"
21
+ licence_statement: "The archive states: \"the contents of the Archive are the intellectual property of their original creators\". Material that includes a distribution license (such as open-source or Creative Commons) is distributed under its licensed terms; \"Any material with no attached license is presumed to be licensed for personal use only.\" The contents may not be compiled for commercial distribution."
22
+ per_collection: null
23
+ training_ok: "no"
24
+ redistribution_ok: "no"
25
+ attribution: "Metadata, indexes and file descriptions (source_meta.description) are released by the archive under CC BY 4.0; attribute the Interactive Fiction Technology Foundation."
26
+ per_file: "Rows whose file description mentions a license carry license_override, with the passage quoted in notes; these passages were not checked one by one."
27
+ licence_tiers: {no/no: 23666, unknown/unknown: 153}
28
+ spdx_top: {"(none)": 23819}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 23819
34
+ units: {file: 23819}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 53269291427
38
+ rows_without_size: 0
39
+ collections:
40
+ distinct: 25
41
+ top:
42
+ games: 19189
43
+ info: 292
44
+ infocom: 1307
45
+ level9: 77
46
+ magazines: 350
47
+ phoenix: 95
48
+ programming: 1019
49
+ rec.arts.int-fiction: 249
50
+ rec.games.int-fiction: 240
51
+ solutions: 663
52
+ sub_collections: null
53
+ batches: null
54
+ n_ext: 382
55
+ usable_unit: null
56
+ form:
57
+ by_ext:
58
+ zip: 5564
59
+ png: 2683
60
+ txt: 1588
61
+ jpg: 1241
62
+ z5: 826
63
+ html: 774
64
+ "(none)": 726
65
+ gz: 683
66
+ js: 623
67
+ dll: 573
68
+ mp3: 537
69
+ gblorb: 530
70
+ pdf: 509
71
+ tgz: 422
72
+ zblorb: 342
73
+ ogg: 317
74
+ css: 309
75
+ exe: 290
76
+ gif: 278
77
+ assets: 267
78
+ "(362 other types)": 4737
79
+ bytes_p50: 83671
80
+ bytes_p95: 4255263
81
+ bytes_max: 1580372181
82
+ size_quantiles:
83
+ max: 1580372181
84
+ min: 0
85
+ p25: 11816
86
+ p50: 83671
87
+ p75: 379176
88
+ p90: 1454631
89
+ p95: 4255263
90
+ p99: 36954632
91
+ p99.9: 230140903
92
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
93
+ by_ext_size:
94
+ - {ext: zip, rows: 5564, bytes: 36455665830, p50: 174840, p95: 19940417, max: 1580372181}
95
+ - {ext: png, rows: 2683, bytes: 793091733, p50: 17335, p95: 1119899, max: 14243107}
96
+ - {ext: txt, rows: 1588, bytes: 78278472, p50: 3466, p95: 414508, max: 1050683}
97
+ - {ext: jpg, rows: 1241, bytes: 318623338, p50: 76944, p95: 971767, max: 14165626}
98
+ - {ext: z5, rows: 826, bytes: 99864217, p50: 114688, p95: 228864, max: 262144}
99
+ - {ext: html, rows: 774, bytes: 292595716, p50: 16925, p95: 1279806, max: 37796142}
100
+ - {ext: "(none)", rows: 726, bytes: 238783880, p50: 32142, p95: 1022404, max: 33761592}
101
+ - {ext: gz, rows: 683, bytes: 3590363057, p50: 42605, p95: 34675696, max: 194934648}
102
+ - {ext: js, rows: 623, bytes: 140766367, p50: 86659, p95: 1304321, max: 5644352}
103
+ - {ext: dll, rows: 573, bytes: 254160136, p50: 30208, p95: 2114008, max: 25882056}
104
+ - {ext: mp3, rows: 537, bytes: 823279205, p50: 698219, p95: 6176182, max: 36979994}
105
+ - {ext: gblorb, rows: 530, bytes: 1939594730, p50: 1182084, p95: 7464444, max: 417101284}
106
+ - {ext: pdf, rows: 509, bytes: 3162665760, p50: 538406, p95: 39266939, max: 120159643}
107
+ - {ext: tgz, rows: 422, bytes: 151012022, p50: 74489, p95: 1044955, max: 14116920}
108
+ - {ext: zblorb, rows: 342, bytes: 164979614, p50: 400000, p95: 1131672, max: 2723610}
109
+ - {ext: ogg, rows: 317, bytes: 249537492, p50: 338499, p95: 3289450, max: 8095525}
110
+ - {ext: css, rows: 309, bytes: 3555829, p50: 3729, p95: 47654, max: 540779}
111
+ - {ext: exe, rows: 290, bytes: 1445328650, p50: 624056, p95: 20474392, max: 210698476}
112
+ - {ext: gif, rows: 278, bytes: 68104383, p50: 46546, p95: 1344893, max: 4211659}
113
+ - {ext: assets, rows: 267, bytes: 49649697, p50: 4376, p95: 725332, max: 9892056}
114
+ over_cap:
115
+ cap: 61944279
116
+ rows_over: 150
117
+ bytes_over: 30056564587
118
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
119
+ largest:
120
+ - {pointer_id: "if-archive:games:competition2020/IFComp2020.zip", size_bytes: 1580372181}
121
+ - {pointer_id: "if-archive:games:competition2026/IFComp2026.zip", size_bytes: 1547371510}
122
+ - {pointer_id: "if-archive:games:competition2025/IFComp2025.zip", size_bytes: 1426121325}
123
+ - {pointer_id: "if-archive:games:competition2024/IFComp2024.zip", size_bytes: 1376263663}
124
+ - {pointer_id: "if-archive:games:springthing/2021/SpringThing2021.zip", size_bytes: 1056261077}
125
+ date_years: null
126
+ media_type: {"(none)": 23819}
127
+ facets:
128
+ content_warning:
129
+ self-harm: 10
130
+ "self-harm, eating-disorder": 1
131
+ text-porn: 37
132
+ "text-porn, self-harm": 3
133
+ unknown: 17
134
+ visual-gore: 9
135
+ kind:
136
+ document: 3372
137
+ game: 2601
138
+ game-package: 8845
139
+ game-source: 488
140
+ interpreter: 425
141
+ media: 5324
142
+ other: 792
143
+ solution: 894
144
+ tool: 1078
145
+ story_format:
146
+ adrift: 165
147
+ agt: 2
148
+ alan2: 13
149
+ alan3: 8
150
+ blorb: 61
151
+ glulx: 591
152
+ hugo: 29
153
+ tads2: 209
154
+ tads3: 105
155
+ zcode: 1418
156
+ artifact: null
157
+ source_origin: null
158
+ parse_cost: null
159
+ parse_quirks: null
160
+ structure:
161
+ self_describing: null
162
+ integrity_builtin: null
163
+ has_reference_output: null
164
+ invariants: null
165
+ tools: null
166
+ ships:
167
+ problem_statement: {present: null}
168
+ reference_solution: {share: null}
169
+ tests: {present: null}
170
+ env_definition: {present: null}
171
+ known_answer: {present: null}
172
+ generator: {present: null}
173
+ upstream_labels: {present: null}
174
+ class: null
175
+ subject:
176
+ domain_span: null
177
+ shape_span: null
178
+ skill_family_span: null
179
+ ont_domain: null
180
+ tech_vocabulary: null
181
+ risk:
182
+ contamination: none
183
+ overlap_check:
184
+ done: true
185
+ pointers_flagged: 0
186
+ pointers_removed: 0
187
+ by_class: {}
188
+ benches: {}
189
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
190
+ pii: null
191
+ pii_evidence: null
192
+ malware: null
193
+ ship_bytes: "no"
194
+ subset:
195
+ rule: "take all rows; licence, size and version are labels, not filters"
196
+ rows_in_subset: 23819
197
+ last_built: null
198
+ access:
199
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
200
+ verify: "Compare size_bytes, md5 and sha512 (published)."
201
+ hosts: {"https://ifarchive.org": 23819}
202
+ uri_example: "https://ifarchive.org/if-archive/art/if-artshow/concept.txt"
203
+ needs_auth: {"false": 23819}
204
+ credential: null
205
+ hash_source: {source_published: 23819}
206
+ range_supported: null
207
+ reachable:
208
+ index_flag: {unknown: 23819}
209
+ index_check: null
210
+ our_sample: null
211
+ drift: null
212
+ sample: null
213
+ evidence:
214
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ class: {fill: unfilled, basis: "ships not judged"}
218
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ form.by_ext:
220
+ fill: auto
221
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
222
+ checked_at: "2026-10-03"
223
+ form.date_years:
224
+ fill: auto
225
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
226
+ checked_at: "2026-10-03"
227
+ form.over_cap:
228
+ fill: auto
229
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
230
+ checked_at: "2026-10-03"
231
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
233
+ form.size_quantiles:
234
+ fill: auto
235
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
236
+ checked_at: "2026-10-03"
237
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
238
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
239
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
240
+ provenance.licence_tiers:
241
+ fill: auto
242
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
243
+ checked_at: "2026-10-03"
244
+ provenance.spdx_top:
245
+ fill: auto
246
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
247
+ checked_at: "2026-10-03"
248
+ risk.contamination:
249
+ fill: auto
250
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
251
+ checked_at: "2026-10-03"
252
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
253
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
256
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
257
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
260
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
261
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
262
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
265
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
266
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
268
+ snapshot.rev:
269
+ fill: auto
270
+ basis: "last commit listed in the publish receipt of this source"
271
+ checked_at: "2026-10-03"
272
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
273
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
275
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ subset:
283
+ fill: auto
284
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
285
+ checked_at: "2026-10-03"
286
+ unit.pointer_count:
287
+ fill: auto
288
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
289
+ checked_at: "2026-10-03"
290
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
292
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/internet-archive-open-licensed.yaml ADDED
@@ -0,0 +1,319 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: internet-archive-open-licensed
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers: {conditional/conditional: 640, conditional/no: 5184, unknown/unknown: 540151, yes/yes: 10264}
29
+ spdx_top:
30
+ "(none)": 540151
31
+ CC-BY-2.5: 1
32
+ CC-BY-3.0: 14
33
+ CC-BY-4.0: 912
34
+ CC-BY-NC-3.0: 1
35
+ CC-BY-NC-4.0: 75
36
+ CC-BY-NC-ND-3.0: 37
37
+ CC-BY-NC-ND-4.0: 1294
38
+ CC-BY-NC-SA-2.5: 2
39
+ CC-BY-NC-SA-3.0: 1
40
+ CC-BY-NC-SA-4.0: 200
41
+ CC-BY-ND-3.0: 49
42
+ CC-BY-ND-4.0: 3804
43
+ CC-BY-SA-3.0: 5
44
+ CC-BY-SA-4.0: 356
45
+ spdx_distinct: 18
46
+ licence_sample_check: null
47
+ unit:
48
+ pointer_level: null
49
+ pointer_count: 556239
50
+ units: {file: 556239}
51
+ state: null
52
+ expanded_count: null
53
+ bytes_total: 87271716895958
54
+ rows_without_size: 0
55
+ collections: null
56
+ sub_collections: null
57
+ batches: null
58
+ n_ext: 2378
59
+ usable_unit: null
60
+ form:
61
+ by_ext:
62
+ js: 137021
63
+ jpg: 69390
64
+ zip: 63439
65
+ png: 51169
66
+ gif: 14151
67
+ json: 11148
68
+ iso: 11015
69
+ "7z": 10778
70
+ ts: 7043
71
+ txt: 6069
72
+ svf: 6004
73
+ rar: 5685
74
+ bin: 5418
75
+ md: 5320
76
+ dsk: 5291
77
+ cia: 4635
78
+ ipa: 4403
79
+ "(none)": 4331
80
+ exe: 4301
81
+ gba: 4052
82
+ "(2358 other types)": 125576
83
+ bytes_p50: 48451
84
+ bytes_p95: 471567933
85
+ bytes_max: 1061063434488
86
+ size_quantiles:
87
+ max: 1061063434488
88
+ min: 0
89
+ p25: 1043
90
+ p50: 48451
91
+ p75: 1857222
92
+ p90: 162961043
93
+ p95: 471567933
94
+ p99: 3091660800
95
+ p99.9: 14442143530
96
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
97
+ by_ext_size:
98
+ - {ext: js, rows: 137021, bytes: 399179506, p50: 361, p95: 5780, max: 7137169}
99
+ - {ext: jpg, rows: 69390, bytes: 30669591518, p50: 88242, p95: 1336105, max: 106124941}
100
+ - {ext: zip, rows: 63439, bytes: 21333895213214, p50: 3681907, p95: 773547190, max: 222989963647}
101
+ - {ext: png, rows: 51169, bytes: 43300998204, p50: 59433, p95: 2396989, max: 281070327}
102
+ - {ext: gif, rows: 14151, bytes: 2544070325, p50: 9733, p95: 1029106, max: 49863844}
103
+ - {ext: json, rows: 11148, bytes: 4775917084, p50: 2193, p95: 1937950, max: 31224701}
104
+ - {ext: iso, rows: 11015, bytes: 18584492033657, p50: 678428672, p95: 5282525184, max: 60649226240}
105
+ - {ext: "7z", rows: 10778, bytes: 6185783488750, p50: 2214833, p95: 3292755535, max: 90710471294}
106
+ - {ext: ts, rows: 7043, bytes: 146812752804, p50: 267, p95: 77621, max: 1990871120}
107
+ - {ext: txt, rows: 6069, bytes: 8340766353, p50: 2037, p95: 398561, max: 1698353728}
108
+ - {ext: svf, rows: 6004, bytes: 923286109060, p50: 112597040, p95: 421898848, max: 2320662360}
109
+ - {ext: rar, rows: 5685, bytes: 6266172024986, p50: 104857600, p95: 4113703979, max: 152356846839}
110
+ - {ext: bin, rows: 5418, bytes: 1136536539963, p50: 2097152, p95: 772375632, max: 62914560512}
111
+ - {ext: md, rows: 5320, bytes: 1000004988, p50: 2637, p95: 1048576, max: 5242880}
112
+ - {ext: dsk, rows: 5291, bytes: 35600376359, p50: 143360, p95: 143360, max: 9283914576}
113
+ - {ext: cia, rows: 4635, bytes: 1137358974076, p50: 27089216, p95: 1585153344, max: 3908690944}
114
+ - {ext: ipa, rows: 4403, bytes: 457850512623, p50: 20878608, p95: 384321623, max: 3496704849}
115
+ - {ext: "(none)", rows: 4331, bytes: 188530590963, p50: 1109, p95: 170459136, max: 6619149668}
116
+ - {ext: exe, rows: 4301, bytes: 766252899505, p50: 35862096, p95: 599177712, max: 24388245928}
117
+ - {ext: gba, rows: 4052, bytes: 36232578194, p50: 8388608, p95: 16777216, max: 33554432}
118
+ over_cap:
119
+ cap: 61944279
120
+ rows_over: 72703
121
+ bytes_over: 86253098716304
122
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
123
+ largest:
124
+ - pointer_id: "internet-archive-open-licensed:software:2024-sony-playstation-hearto-1g1r-collection/2024-sony-playstation-hearto-1g1r-collection_resume.tar.gz"
125
+ size_bytes: 1061063434488
126
+ - pointer_id: "internet-archive-open-licensed:software:2-tb-xbox-origins-xbox-jl/[2TB]-Xbox_Origins-[XBOX]-JL_resume.tar.gz"
127
+ size_bytes: 1033289249277
128
+ - pointer_id: "internet-archive-open-licensed:software:188.227.23.100files_iso-ftp-archive/188.227.23.100_files_iso.tar.gz"
129
+ size_bytes: 914436821658
130
+ - pointer_id: "internet-archive-open-licensed:software:2016.cyanogenmod.org.snapshot/cyanogenmod.org.snapshop.tar"
131
+ size_bytes: 692793989120
132
+ - pointer_id: "internet-archive-open-licensed:software:2015.07.ftp.luth.se/ftp.luth.se_NetBSD.tar"
133
+ size_bytes: 665534484480
134
+ date_years: null
135
+ media_type: {"(none)": 556239}
136
+ facets:
137
+ emulator:
138
+ a2600: 273
139
+ a800xlp: 122
140
+ apple2e: 148
141
+ apple2ee: 49335
142
+ apple2ee-helper: 120
143
+ coco3disk: 74
144
+ cpc6128: 192
145
+ dosbox: 1822
146
+ gba: 134
147
+ gbcolor: 94
148
+ genesis: 1130
149
+ megadriv: 155
150
+ nes: 395
151
+ ruffle-swf: 186950
152
+ sae-a500p: 2836
153
+ snes: 90
154
+ tg16: 85
155
+ vice: 244
156
+ vice-resid: 24957
157
+ zx81: 187
158
+ emulator_ext:
159
+ C10: 34
160
+ ZIP: 345
161
+ adf: 2836
162
+ atr: 141
163
+ bin: 1400
164
+ cartridge: 73
165
+ d64: 25195
166
+ dsk: 49850
167
+ gb: 36
168
+ gba: 127
169
+ gbc: 36
170
+ img: 53
171
+ md: 161
172
+ nes: 409
173
+ nib: 34
174
+ p: 187
175
+ pce: 85
176
+ st: 54
177
+ swf: 186942
178
+ zip: 1577
179
+ licenseurl:
180
+ "https://opensource.org/license/mit": 3
181
+ "https://opensource.org/licenses/MIT": 35
182
+ "https://unlicense.org/UNLICENSE": 3
183
+ artifact: null
184
+ source_origin: null
185
+ parse_cost: null
186
+ parse_quirks: null
187
+ structure:
188
+ self_describing: null
189
+ integrity_builtin: null
190
+ has_reference_output: null
191
+ invariants: null
192
+ tools: null
193
+ ships:
194
+ problem_statement: {present: null}
195
+ reference_solution: {share: null}
196
+ tests: {present: null}
197
+ env_definition: {present: null}
198
+ known_answer: {present: null}
199
+ generator: {present: null}
200
+ upstream_labels: {present: null}
201
+ class: null
202
+ subject:
203
+ domain_span: null
204
+ shape_span: null
205
+ skill_family_span: null
206
+ ont_domain: null
207
+ tech_vocabulary: null
208
+ risk:
209
+ contamination: none
210
+ overlap_check:
211
+ done: true
212
+ pointers_flagged: 22
213
+ pointers_removed: 0
214
+ by_class: {generic_text: 14, trivial_small: 8}
215
+ benches: {"Agents' Last Exam": 4, "FrontierSWE v2": 19, NL2Repo: 3, Terminal-Bench: 1}
216
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
217
+ pii: null
218
+ pii_evidence: null
219
+ malware: null
220
+ ship_bytes: unknown
221
+ subset:
222
+ rule: "take all rows; licence, size and version are labels, not filters"
223
+ rows_in_subset: 556239
224
+ last_built: null
225
+ access:
226
+ method: null
227
+ verify: null
228
+ hosts: {"https://archive.org": 556239}
229
+ uri_example: "https://archive.org/download/0-0-n-0-f-413-3-n-master/%240%240%20N0%20F413%233N_Master.m3u"
230
+ needs_auth: {"false": 556239}
231
+ credential: null
232
+ hash_source: {source_published: 556239}
233
+ range_supported: null
234
+ reachable:
235
+ index_flag: {unknown: 556239}
236
+ index_check: null
237
+ our_sample: null
238
+ drift: null
239
+ sample: null
240
+ evidence:
241
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
242
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
243
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
244
+ class: {fill: unfilled, basis: "ships not judged"}
245
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ form.by_ext:
247
+ fill: auto
248
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
249
+ checked_at: "2026-10-03"
250
+ form.date_years:
251
+ fill: auto
252
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
253
+ checked_at: "2026-10-03"
254
+ form.over_cap:
255
+ fill: auto
256
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
257
+ checked_at: "2026-10-03"
258
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
260
+ form.size_quantiles:
261
+ fill: auto
262
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
263
+ checked_at: "2026-10-03"
264
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
265
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
266
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ provenance.licence_tiers:
268
+ fill: auto
269
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
270
+ checked_at: "2026-10-03"
271
+ provenance.spdx_top:
272
+ fill: auto
273
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
274
+ checked_at: "2026-10-03"
275
+ risk.contamination:
276
+ fill: auto
277
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
278
+ checked_at: "2026-10-03"
279
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
283
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
287
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
290
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
293
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
294
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
295
+ snapshot.rev:
296
+ fill: auto
297
+ basis: "last commit listed in the publish receipt of this source"
298
+ checked_at: "2026-10-03"
299
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
300
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
301
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
302
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
303
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
304
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
305
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
306
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
307
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
308
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
309
+ subset:
310
+ fill: auto
311
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
312
+ checked_at: "2026-10-03"
313
+ unit.pointer_count:
314
+ fill: auto
315
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
316
+ checked_at: "2026-10-03"
317
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
318
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
319
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/kde-webkit-gentoo-bugzilla.yaml ADDED
@@ -0,0 +1,331 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: kde-webkit-gentoo-bugzilla
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers: {unknown/unknown: 892609}
29
+ spdx_top: {"(none)": 892609}
30
+ spdx_distinct: 1
31
+ licence_sample_check: null
32
+ unit:
33
+ pointer_level: null
34
+ pointer_count: 892609
35
+ units: {"API response": 892609}
36
+ state: null
37
+ expanded_count: null
38
+ bytes_total: 50473548252
39
+ rows_without_size: 0
40
+ collections:
41
+ distinct: 2
42
+ top: {attachment: 270670, bug: 621939}
43
+ sub_collections: null
44
+ batches: null
45
+ n_ext: 1817
46
+ usable_unit: null
47
+ form:
48
+ by_ext:
49
+ json: 622047
50
+ png: 72658
51
+ ebuild: 31490
52
+ patch: 26507
53
+ "(none)": 25467
54
+ txt: 20193
55
+ diff: 11208
56
+ jpg: 9844
57
+ html: 7576
58
+ log: 7138
59
+ kcrash: 6249
60
+ mp4: 4570
61
+ gz: 4021
62
+ jpeg: 3428
63
+ xml: 3041
64
+ webm: 2901
65
+ bz2: 2373
66
+ pdf: 1783
67
+ zip: 1720
68
+ mkv: 1710
69
+ "(1797 other types)": 26685
70
+ bytes_p50: 2406
71
+ bytes_p95: 180695
72
+ bytes_max: 8816973
73
+ size_quantiles:
74
+ max: 8816973
75
+ min: 0
76
+ p25: 987
77
+ p50: 2406
78
+ p75: 8664
79
+ p90: 54771
80
+ p95: 180695
81
+ p99: 1522222
82
+ p99.9: 3740396
83
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
84
+ by_ext_size:
85
+ - {ext: json, rows: 622047, bytes: 2744764608, p50: 1817, p95: 16553, max: 2720382}
86
+ - {ext: png, rows: 72658, bytes: 16419676695, p50: 88445, p95: 935302, max: 4095501}
87
+ - {ext: ebuild, rows: 31490, bytes: 53165023, p50: 1037, p95: 4950, max: 86105}
88
+ - {ext: patch, rows: 26507, bytes: 343564307, p50: 1369, p95: 28485, max: 3756665}
89
+ - {ext: "(none)", rows: 25467, bytes: 930496275, p50: 6428, p95: 101909, max: 4096000}
90
+ - {ext: txt, rows: 20193, bytes: 1284214891, p50: 10038, p95: 233107, max: 4073419}
91
+ - {ext: diff, rows: 11208, bytes: 104227840, p50: 1290, p95: 24150, max: 3929247}
92
+ - {ext: jpg, rows: 9844, bytes: 3985446334, p50: 121402, p95: 2354525, max: 4084428}
93
+ - {ext: html, rows: 7576, bytes: 52068702, p50: 1608, p95: 13723, max: 2661267}
94
+ - {ext: log, rows: 7138, bytes: 1224080432, p50: 35579, p95: 830927, max: 4083697}
95
+ - {ext: kcrash, rows: 6249, bytes: 284289697, p50: 9535, p95: 185890, max: 1249799}
96
+ - {ext: mp4, rows: 4570, bytes: 7701444017, p50: 1445543, p95: 3840001, max: 4096000}
97
+ - {ext: gz, rows: 4021, bytes: 603917408, p50: 5336, p95: 792815, max: 4012926}
98
+ - {ext: jpeg, rows: 3428, bytes: 645321594, p50: 129415, p95: 493734, max: 3896578}
99
+ - {ext: xml, rows: 3041, bytes: 93747414, p50: 12128, p95: 65959, max: 4093586}
100
+ - {ext: webm, rows: 2901, bytes: 4022503663, p50: 1107621, p95: 3601125, max: 4093638}
101
+ - {ext: bz2, rows: 2373, bytes: 240119766, p50: 6975, p95: 527008, max: 4075520}
102
+ - {ext: pdf, rows: 1783, bytes: 627157266, p50: 109452, p95: 1560724, max: 3969092}
103
+ - {ext: zip, rows: 1720, bytes: 823453356, p50: 101268, p95: 2488518, max: 4050335}
104
+ - {ext: mkv, rows: 1710, bytes: 2983665506, p50: 1547260, p95: 3784888, max: 4092567}
105
+ over_cap:
106
+ cap: 61944279
107
+ rows_over: 0
108
+ bytes_over: 0
109
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
110
+ largest:
111
+ - {pointer_id: "kde-webkit-gentoo-bugzilla:attachment:kde-26583", size_bytes: 8816973}
112
+ - {pointer_id: "kde-webkit-gentoo-bugzilla:attachment:kde-21937", size_bytes: 8361550}
113
+ - {pointer_id: "kde-webkit-gentoo-bugzilla:attachment:kde-185782", size_bytes: 4096000}
114
+ - {pointer_id: "kde-webkit-gentoo-bugzilla:attachment:kde-185781", size_bytes: 4096000}
115
+ - {pointer_id: "kde-webkit-gentoo-bugzilla:attachment:kde-185780", size_bytes: 4096000}
116
+ date_years: null
117
+ media_type:
118
+ application/json: 622038
119
+ application/octet-stream: 11118
120
+ image/jpeg: 12468
121
+ image/png: 71339
122
+ text/html: 7554
123
+ text/plain: 130718
124
+ text/x-log: 2687
125
+ video/mp4: 4638
126
+ video/webm: 2896
127
+ video/x-matroska: 1682
128
+ facets:
129
+ attachment_category: {config: 3578, crash: 29153, log: 9739, other: 186436, patch: 41764}
130
+ is_obsolete: {"false": 246179, "true": 24491}
131
+ is_patch: {"false": 231298, "true": 39372}
132
+ op_sys:
133
+ All: 27806
134
+ "Android 10.x": 105
135
+ "Android 11.x": 146
136
+ "Android 12.x": 70
137
+ "Android 13.x": 208
138
+ "Android 14.x": 236
139
+ "Android 4.x": 111
140
+ FreeBSD: 3385
141
+ IRIX: 144
142
+ Linux: 537884
143
+ "Microsoft Windows": 13895
144
+ "Microsoft Windows CE": 202
145
+ NetBSD: 254
146
+ "OS X": 162
147
+ OpenBSD: 137
148
+ Other: 22358
149
+ Solaris: 1574
150
+ Unspecified: 10019
151
+ iOS: 86
152
+ macOS: 2733
153
+ resolution:
154
+ BACKTRACE: 225
155
+ CANTFIX: 1324
156
+ DOWNSTREAM: 6464
157
+ DUPLICATE: 125945
158
+ FIXED: 261201
159
+ INVALID: 46818
160
+ LATER: 2790
161
+ MOVED: 480
162
+ NEEDINFO: 2116
163
+ OBSOLETE: 177
164
+ PKGREMOVED: 1
165
+ REMIND: 1406
166
+ TEST-REQUEST: 1496
167
+ UNMAINTAINED: 27997
168
+ UPSTREAM: 12389
169
+ WAITINGFORINFO: 3676
170
+ WONTFIX: 21229
171
+ WORKSFORME: 65379
172
+ severity:
173
+ blocker: 2925
174
+ crash: 125153
175
+ critical: 7092
176
+ enhancement: 24877
177
+ grave: 5454
178
+ major: 20956
179
+ minor: 24307
180
+ normal: 326465
181
+ task: 625
182
+ trivial: 2437
183
+ wishlist: 81648
184
+ site: {gentoo: 180646, kde: 711963}
185
+ status:
186
+ ASSIGNED: 1144
187
+ CLOSED: 40067
188
+ CONFIRMED: 14110
189
+ IN_PROGRESS: 6
190
+ NEEDSINFO: 216
191
+ REOPENED: 1170
192
+ RESOLVED: 535002
193
+ UNCONFIRMED: 24396
194
+ VERIFIED: 5828
195
+ artifact: null
196
+ source_origin: null
197
+ parse_cost: null
198
+ parse_quirks: null
199
+ structure:
200
+ self_describing: null
201
+ integrity_builtin: null
202
+ has_reference_output: null
203
+ invariants: null
204
+ tools: null
205
+ ships:
206
+ problem_statement: {present: null}
207
+ reference_solution: {share: null}
208
+ tests: {present: null}
209
+ env_definition: {present: null}
210
+ known_answer: {present: null}
211
+ generator: {present: null}
212
+ upstream_labels: {present: null}
213
+ class: null
214
+ subject:
215
+ domain_span: null
216
+ shape_span: null
217
+ skill_family_span: null
218
+ ont_domain: null
219
+ tech_vocabulary: null
220
+ risk:
221
+ contamination: none
222
+ overlap_check:
223
+ done: true
224
+ pointers_flagged: 0
225
+ pointers_removed: 0
226
+ by_class: {}
227
+ benches: {}
228
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
229
+ pii: null
230
+ pii_evidence: null
231
+ malware: null
232
+ ship_bytes: unknown
233
+ subset:
234
+ rule: "take all rows; licence, size and version are labels, not filters"
235
+ rows_in_subset: 892609
236
+ last_built: null
237
+ access:
238
+ method: null
239
+ verify: null
240
+ hosts: {"https://bugs.gentoo.org": 180646, "https://bugs.kde.org": 711963}
241
+ uri_example: "https://bugs.gentoo.org/rest/bug/attachment/1"
242
+ needs_auth: {"false": 892609}
243
+ credential: null
244
+ hash_source: {computed: 621939, computed_on_fetch: 270670}
245
+ range_supported: null
246
+ reachable:
247
+ index_flag: {unknown: 892609}
248
+ index_check: null
249
+ our_sample: null
250
+ drift: null
251
+ sample: null
252
+ evidence:
253
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ class: {fill: unfilled, basis: "ships not judged"}
257
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ form.by_ext:
259
+ fill: auto
260
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
261
+ checked_at: "2026-10-03"
262
+ form.date_years:
263
+ fill: auto
264
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
265
+ checked_at: "2026-10-03"
266
+ form.over_cap:
267
+ fill: auto
268
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
269
+ checked_at: "2026-10-03"
270
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
272
+ form.size_quantiles:
273
+ fill: auto
274
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
275
+ checked_at: "2026-10-03"
276
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
278
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ provenance.licence_tiers:
280
+ fill: auto
281
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
282
+ checked_at: "2026-10-03"
283
+ provenance.spdx_top:
284
+ fill: auto
285
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
286
+ checked_at: "2026-10-03"
287
+ risk.contamination:
288
+ fill: auto
289
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
290
+ checked_at: "2026-10-03"
291
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
293
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
294
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
295
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
296
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
297
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
298
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
299
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
300
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
301
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
302
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
303
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
304
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
305
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
306
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
307
+ snapshot.rev:
308
+ fill: auto
309
+ basis: "last commit listed in the publish receipt of this source"
310
+ checked_at: "2026-10-03"
311
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
312
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
313
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
314
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
315
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
316
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
317
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
318
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
319
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
320
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
321
+ subset:
322
+ fill: auto
323
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
324
+ checked_at: "2026-10-03"
325
+ unit.pointer_count:
326
+ fill: auto
327
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
328
+ checked_at: "2026-10-03"
329
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
330
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
331
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/kernel-bugzilla.yaml ADDED
@@ -0,0 +1,318 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: kernel-bugzilla
8
+ parent: null
9
+ title: "Linux kernel Bugzilla bugs and attachments"
10
+ description: "All publicly visible bugs and public attachments on the Linux kernel Bugzilla (bugzilla.kernel.org), listed through the Bugzilla REST API in bug number order. One pointer per bug (collection bug) and one per public attachment (collection attachment). Attachments are mostly what kernel bug reports carry: dmesg and other kernel logs, kernel configs, patches, crash traces and hardware dumps. For a bug the hash covers its comment list as returned by the API; an attachment is retrieved through the REST API, which returns its content base64 encoded. Product, component, kernel version, status, severity, regression flag, keywords, attachment media type and an attachment category are in source_meta."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-02"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Copyright of the contributors; no license stated"
20
+ licence_url: "https://www.kernel.org/legal.html"
21
+ licence_statement: "Bugs, comments and attachments remain the copyright of their submitters, and Bugzilla pages state no license. The kernel.org legal page reads: \"Except where otherwise stated, content on this site is copyright (C) 1997-2014 by The Linux Kernel Organization, Inc. and is made available to you under the Creative Commons Attribution ShareAlike 4.0 International License. Distributed software is copyrighted by their respective contributors and are distributed under their own individual licenses.\" It does not say whether this covers content submitted to Bugzilla by users. Patches are likely GPL-2.0 by kernel convention (noted per row in source_meta.license_hint, not checked)."
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {unknown/unknown: 125846}
28
+ spdx_top: {"(none)": 125846}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 125846
34
+ units: {"API response": 125846}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 10574409053
38
+ rows_without_size: 0
39
+ collections:
40
+ distinct: 2
41
+ top: {attachment: 81259, bug: 44587}
42
+ sub_collections: null
43
+ batches: null
44
+ n_ext: 1435
45
+ usable_unit: null
46
+ form:
47
+ by_ext:
48
+ json: 44590
49
+ txt: 19955
50
+ "(none)": 18188
51
+ patch: 10020
52
+ log: 6715
53
+ dat: 5969
54
+ gz: 2064
55
+ jpg: 1833
56
+ diff: 1667
57
+ html: 1211
58
+ png: 1111
59
+ out: 1077
60
+ c: 947
61
+ zip: 788
62
+ bz2: 710
63
+ dsl: 479
64
+ xz: 400
65
+ dmesg: 345
66
+ sh: 284
67
+ config: 249
68
+ "(1415 other types)": 7244
69
+ bytes_p50: 5965
70
+ bytes_p95: 289882
71
+ bytes_max: 13154512
72
+ size_quantiles:
73
+ max: 13154512
74
+ min: 0
75
+ p25: 1672
76
+ p50: 5965
77
+ p75: 41158
78
+ p90: 127695
79
+ p95: 289882
80
+ p99: 1735279
81
+ p99.9: 4500670
82
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
83
+ by_ext_size:
84
+ - {ext: json, rows: 44590, bytes: 342594722, p50: 2870, p95: 28370, max: 13154512}
85
+ - {ext: txt, rows: 19955, bytes: 2098105955, p50: 29346, p95: 362545, max: 5242880}
86
+ - {ext: "(none)", rows: 18188, bytes: 1362294042, p50: 27081, p95: 237904, max: 5242880}
87
+ - {ext: patch, rows: 10020, bytes: 39378204, p50: 1772, p95: 11467, max: 328830}
88
+ - {ext: log, rows: 6715, bytes: 936229231, p50: 51170, p95: 497119, max: 5242820}
89
+ - {ext: dat, rows: 5969, bytes: 329085127, p50: 4136, p95: 4676, max: 5017600}
90
+ - {ext: gz, rows: 2064, bytes: 439666207, p50: 30221, p95: 1006643, max: 4957127}
91
+ - {ext: jpg, rows: 1833, bytes: 1512541068, p50: 399719, p95: 3394416, max: 5218650}
92
+ - {ext: diff, rows: 1667, bytes: 8740849, p50: 1274, p95: 15094, max: 541780}
93
+ - {ext: html, rows: 1211, bytes: 125447664, p50: 4684, p95: 728049, max: 4184527}
94
+ - {ext: png, rows: 1111, bytes: 294303245, p50: 93084, p95: 983007, max: 5156914}
95
+ - {ext: out, rows: 1077, bytes: 183780520, p50: 39731, p95: 798163, max: 5005892}
96
+ - {ext: c, rows: 947, bytes: 13270508, p50: 2828, p95: 69419, max: 277379}
97
+ - {ext: zip, rows: 788, bytes: 294295612, p50: 70943, p95: 2325537, max: 5150637}
98
+ - {ext: bz2, rows: 710, bytes: 182220744, p50: 31515, p95: 927585, max: 4957582}
99
+ - {ext: dsl, rows: 479, bytes: 238984659, p50: 301459, p95: 1864723, max: 4015482}
100
+ - {ext: xz, rows: 400, bytes: 224907468, p50: 61544, p95: 3324976, max: 5043120}
101
+ - {ext: dmesg, rows: 345, bytes: 25179371, p50: 56126, p95: 197479, max: 1214126}
102
+ - {ext: sh, rows: 284, bytes: 948860, p50: 888, p95: 8465, max: 280293}
103
+ - {ext: config, rows: 249, bytes: 23696867, p50: 77163, p95: 221665, max: 307383}
104
+ over_cap:
105
+ cap: 61944279
106
+ rows_over: 0
107
+ bytes_over: 0
108
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
109
+ largest:
110
+ - {pointer_id: "kernel-bugzilla:bug:2369", size_bytes: 13154512}
111
+ - {pointer_id: "kernel-bugzilla:attachment:278587", size_bytes: 5242880}
112
+ - {pointer_id: "kernel-bugzilla:attachment:278585", size_bytes: 5242880}
113
+ - {pointer_id: "kernel-bugzilla:attachment:276959", size_bytes: 5242880}
114
+ - {pointer_id: "kernel-bugzilla:attachment:310443", size_bytes: 5242820}
115
+ date_years: null
116
+ media_type:
117
+ application/gzip: 937
118
+ application/json: 44590
119
+ application/octet-stream: 8024
120
+ image/jpeg: 1904
121
+ image/png: 1112
122
+ message/delivery-status: 2818
123
+ text/html: 1368
124
+ text/plain: 55656
125
+ text/rfc822-headers: 2815
126
+ text/x-log: 1394
127
+ facets:
128
+ category: {config: 4288, log: 33370, other: 29584, patch: 14017}
129
+ is_obsolete: {"false": 75339, "true": 5920}
130
+ is_patch: {"false": 68228, "true": 13031}
131
+ keywords: {opw: 27, "opw, trivial": 6, trivial: 89}
132
+ product:
133
+ ACPI: 3668
134
+ "Alternate Trees": 221
135
+ Documentation: 449
136
+ Drivers: 18389
137
+ EFI: 87
138
+ "File System": 4668
139
+ IO/Storage: 2784
140
+ Linux: 494
141
+ "Memory Management": 1417
142
+ Networking: 2796
143
+ Other: 2415
144
+ "Platform Specific/Hardware": 2364
145
+ "Power Management": 1413
146
+ "Process Management": 577
147
+ "SCSI Drivers": 549
148
+ Timers: 301
149
+ Tools: 535
150
+ Tracing/Profiling: 166
151
+ Virtualization: 573
152
+ v4l-dvb: 680
153
+ regression: {"No": 27331, "Yes": 7229}
154
+ resolution:
155
+ ANSWERED: 609
156
+ CODE_FIX: 11654
157
+ DISTRO_KERNEL: 43
158
+ DOCUMENTED: 852
159
+ DUPLICATE: 1990
160
+ IMPLEMENTED: 67
161
+ INSUFFICIENT_DATA: 2588
162
+ INVALID: 4699
163
+ MOVED: 226
164
+ OBSOLETE: 3276
165
+ PATCH_ALREADY_AVAILABLE: 2804
166
+ TAINTED_KERNEL: 14
167
+ UNREPRODUCIBLE: 1724
168
+ WILL_FIX_LATER: 118
169
+ WILL_NOT_FIX: 1353
170
+ severity: {blocking: 2378, enhancement: 654, high: 6923, low: 1922, normal: 32710}
171
+ status:
172
+ ASSIGNED: 242
173
+ CLOSED: 12979
174
+ DEFERRED: 25
175
+ NEEDINFO: 504
176
+ NEW: 11604
177
+ REJECTED: 5048
178
+ REOPENED: 220
179
+ RESOLVED: 13925
180
+ VERIFIED: 40
181
+ version: {"2.5": 41676, unspecified: 2911}
182
+ artifact: null
183
+ source_origin: null
184
+ parse_cost: null
185
+ parse_quirks: null
186
+ structure:
187
+ self_describing: null
188
+ integrity_builtin: null
189
+ has_reference_output: null
190
+ invariants: null
191
+ tools: null
192
+ ships:
193
+ problem_statement: {present: null}
194
+ reference_solution: {share: null}
195
+ tests: {present: null}
196
+ env_definition: {present: null}
197
+ known_answer: {present: null}
198
+ generator: {present: null}
199
+ upstream_labels: {present: null}
200
+ class: null
201
+ subject:
202
+ domain_span: null
203
+ shape_span: null
204
+ skill_family_span: null
205
+ ont_domain: null
206
+ tech_vocabulary: null
207
+ risk:
208
+ contamination: none
209
+ overlap_check:
210
+ done: true
211
+ pointers_flagged: 0
212
+ pointers_removed: 0
213
+ by_class: {}
214
+ benches: {}
215
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
216
+ pii: null
217
+ pii_evidence: null
218
+ malware: null
219
+ ship_bytes: unknown
220
+ subset:
221
+ rule: "take all rows; licence, size and version are labels, not filters"
222
+ rows_in_subset: 125846
223
+ last_built: null
224
+ access:
225
+ method: "Bug rows and attachment rows: GET uri; the response is JSON."
226
+ verify: "Bug rows: take bugs.<bug number>.comments (the path is in source_meta.api_body_field; a list of objects with a single text key), serialize it with json.dumps(ensure_ascii=False), encode it as UTF-8 and compare size and sha256; the stored file is that JSON. Attachment rows: take attachments.<attachment number>.data (the path is in source_meta.api_body_field), decode it as base64 (source_meta.api_body_decode) and compare the size of the decoded bytes; the stored file is the decoded attachment. The source publishes no hash, so compute it on download."
227
+ hosts: {"https://bugzilla.kernel.org": 125846}
228
+ uri_example: "https://bugzilla.kernel.org/rest/bug/attachment/1"
229
+ needs_auth: {"false": 125846}
230
+ credential: null
231
+ hash_source: {computed: 44587, computed_on_fetch: 81259}
232
+ range_supported: null
233
+ reachable:
234
+ index_flag: {unknown: 125846}
235
+ index_check: null
236
+ our_sample: null
237
+ drift: null
238
+ sample: null
239
+ evidence:
240
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
241
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
242
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
243
+ class: {fill: unfilled, basis: "ships not judged"}
244
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
245
+ form.by_ext:
246
+ fill: auto
247
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
248
+ checked_at: "2026-10-03"
249
+ form.date_years:
250
+ fill: auto
251
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
252
+ checked_at: "2026-10-03"
253
+ form.over_cap:
254
+ fill: auto
255
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
256
+ checked_at: "2026-10-03"
257
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ form.size_quantiles:
260
+ fill: auto
261
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
262
+ checked_at: "2026-10-03"
263
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
265
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
266
+ provenance.licence_tiers:
267
+ fill: auto
268
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
269
+ checked_at: "2026-10-03"
270
+ provenance.spdx_top:
271
+ fill: auto
272
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
273
+ checked_at: "2026-10-03"
274
+ risk.contamination:
275
+ fill: auto
276
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
277
+ checked_at: "2026-10-03"
278
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
282
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
287
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
290
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
293
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
294
+ snapshot.rev:
295
+ fill: auto
296
+ basis: "last commit listed in the publish receipt of this source"
297
+ checked_at: "2026-10-03"
298
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
299
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
300
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
301
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
302
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
303
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
304
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
305
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
306
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
307
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
308
+ subset:
309
+ fill: auto
310
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
311
+ checked_at: "2026-10-03"
312
+ unit.pointer_count:
313
+ fill: auto
314
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
315
+ checked_at: "2026-10-03"
316
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
317
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
318
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/launchpad-ubuntu-bugs.yaml ADDED
@@ -0,0 +1,289 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: launchpad-ubuntu-bugs
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers: {unknown/unknown: 43672}
29
+ spdx_top: {"(none)": 43672}
30
+ spdx_distinct: 1
31
+ licence_sample_check: null
32
+ unit:
33
+ pointer_level: null
34
+ pointer_count: 43672
35
+ units: {"API response": 32812, file: 10860}
36
+ state: null
37
+ expanded_count: null
38
+ bytes_total: 24306758
39
+ rows_without_size: 10860
40
+ collections:
41
+ distinct: 2
42
+ top: {attachment: 10860, bug: 32812}
43
+ sub_collections: null
44
+ batches: null
45
+ n_ext: 167
46
+ usable_unit: null
47
+ form:
48
+ by_ext:
49
+ json: 32812
50
+ "(none)": 9047
51
+ log: 367
52
+ txt: 343
53
+ conf: 227
54
+ debdiff: 132
55
+ png: 127
56
+ patch: 58
57
+ diff: 55
58
+ "1": 35
59
+ sh: 27
60
+ gz: 22
61
+ lst: 21
62
+ jpg: 20
63
+ pdf: 19
64
+ list: 14
65
+ "10": 12
66
+ xml: 12
67
+ c: 10
68
+ out: 10
69
+ "(147 other types)": 302
70
+ bytes_p50: 379
71
+ bytes_p95: 1832
72
+ bytes_max: 51391
73
+ size_quantiles:
74
+ max: 51391
75
+ min: 4
76
+ p25: 181
77
+ p50: 379
78
+ p75: 600
79
+ p90: 1091
80
+ p95: 1832
81
+ p99: 8421
82
+ p99.9: 37252
83
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
84
+ by_ext_size:
85
+ - {ext: json, rows: 32812, bytes: 24306758, p50: 379, p95: 1832, max: 51391}
86
+ - {ext: "(none)", rows: 9047, bytes: 0, p50: null, p95: null, max: null}
87
+ - {ext: log, rows: 367, bytes: 0, p50: null, p95: null, max: null}
88
+ - {ext: txt, rows: 343, bytes: 0, p50: null, p95: null, max: null}
89
+ - {ext: conf, rows: 227, bytes: 0, p50: null, p95: null, max: null}
90
+ - {ext: debdiff, rows: 132, bytes: 0, p50: null, p95: null, max: null}
91
+ - {ext: png, rows: 127, bytes: 0, p50: null, p95: null, max: null}
92
+ - {ext: patch, rows: 58, bytes: 0, p50: null, p95: null, max: null}
93
+ - {ext: diff, rows: 55, bytes: 0, p50: null, p95: null, max: null}
94
+ - {ext: "1", rows: 35, bytes: 0, p50: null, p95: null, max: null}
95
+ - {ext: sh, rows: 27, bytes: 0, p50: null, p95: null, max: null}
96
+ - {ext: gz, rows: 22, bytes: 0, p50: null, p95: null, max: null}
97
+ - {ext: lst, rows: 21, bytes: 0, p50: null, p95: null, max: null}
98
+ - {ext: jpg, rows: 20, bytes: 0, p50: null, p95: null, max: null}
99
+ - {ext: pdf, rows: 19, bytes: 0, p50: null, p95: null, max: null}
100
+ - {ext: list, rows: 14, bytes: 0, p50: null, p95: null, max: null}
101
+ - {ext: "10", rows: 12, bytes: 0, p50: null, p95: null, max: null}
102
+ - {ext: xml, rows: 12, bytes: 0, p50: null, p95: null, max: null}
103
+ - {ext: c, rows: 10, bytes: 0, p50: null, p95: null, max: null}
104
+ - {ext: out, rows: 10, bytes: 0, p50: null, p95: null, max: null}
105
+ over_cap:
106
+ cap: 61944279
107
+ rows_over: 0
108
+ bytes_over: 0
109
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
110
+ largest:
111
+ - {pointer_id: "launchpad-ubuntu-bugs:bug:25192", size_bytes: 51391}
112
+ - {pointer_id: "launchpad-ubuntu-bugs:bug:8389", size_bytes: 50009}
113
+ - {pointer_id: "launchpad-ubuntu-bugs:bug:11805", size_bytes: 50005}
114
+ - {pointer_id: "launchpad-ubuntu-bugs:bug:13944", size_bytes: 50004}
115
+ - {pointer_id: "launchpad-ubuntu-bugs:bug:33844", size_bytes: 50001}
116
+ date_years: null
117
+ media_type: {"(none)": 10860, application/json: 32812}
118
+ facets:
119
+ apport: {"false": 32803, "true": 9}
120
+ attachment_category:
121
+ config: 461
122
+ coredump: 2
123
+ dependencies: 3
124
+ log: 1744
125
+ other: 5905
126
+ patch: 2435
127
+ procmaps: 21
128
+ stacktrace: 289
129
+ information_type: {Public: 32797, "Public Security": 15}
130
+ is_patch: {"false": 8451, "true": 2409}
131
+ series:
132
+ Breezy: 55
133
+ "Breezy, Dapper": 14
134
+ Dapper: 122
135
+ "Dapper, Edgy": 3
136
+ Edgy: 2
137
+ Feisty: 2
138
+ Gutsy: 3
139
+ Hardy: 15
140
+ "Hardy, Intrepid": 2
141
+ Hoary: 11
142
+ "Hoary, Breezy": 6
143
+ Intrepid: 4
144
+ Jaunty: 4
145
+ Karmic: 2
146
+ "Linaro Evaluation Build": 1
147
+ Lucid: 6
148
+ "Natty, Oneiric": 1
149
+ "Precise, Quantal": 1
150
+ "Warty, Hoary, Breezy": 3
151
+ "Warty, Hoary, Breezy, Dapper": 2
152
+ type: {Patch: 2409, Unspecified: 8451}
153
+ artifact: null
154
+ source_origin: null
155
+ parse_cost: null
156
+ parse_quirks: null
157
+ structure:
158
+ self_describing: null
159
+ integrity_builtin: null
160
+ has_reference_output: null
161
+ invariants: null
162
+ tools: null
163
+ ships:
164
+ problem_statement: {present: null}
165
+ reference_solution: {share: null}
166
+ tests: {present: null}
167
+ env_definition: {present: null}
168
+ known_answer: {present: null}
169
+ generator: {present: null}
170
+ upstream_labels: {present: null}
171
+ class: null
172
+ subject:
173
+ domain_span: null
174
+ shape_span: null
175
+ skill_family_span: null
176
+ ont_domain: null
177
+ tech_vocabulary: null
178
+ risk:
179
+ contamination: none
180
+ overlap_check:
181
+ done: true
182
+ pointers_flagged: 0
183
+ pointers_removed: 0
184
+ by_class: {}
185
+ benches: {}
186
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
187
+ pii: null
188
+ pii_evidence: null
189
+ malware: null
190
+ ship_bytes: unknown
191
+ subset:
192
+ rule: "take all rows; licence, size and version are labels, not filters"
193
+ rows_in_subset: 43672
194
+ last_built: null
195
+ access:
196
+ method: null
197
+ verify: null
198
+ hosts: {"https://api.launchpad.net": 43672}
199
+ uri_example: "https://api.launchpad.net/devel/bugs/870/+attachment/10/data"
200
+ needs_auth: {"false": 43672}
201
+ credential: null
202
+ hash_source: {computed: 32812, computed_on_fetch: 10860}
203
+ range_supported: null
204
+ reachable:
205
+ index_flag: {unknown: 43672}
206
+ index_check: null
207
+ our_sample: null
208
+ drift: null
209
+ sample: null
210
+ evidence:
211
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
212
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
213
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
214
+ class: {fill: unfilled, basis: "ships not judged"}
215
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ form.by_ext:
217
+ fill: auto
218
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
219
+ checked_at: "2026-10-03"
220
+ form.date_years:
221
+ fill: auto
222
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
223
+ checked_at: "2026-10-03"
224
+ form.over_cap:
225
+ fill: auto
226
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
227
+ checked_at: "2026-10-03"
228
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ form.size_quantiles:
231
+ fill: auto
232
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
233
+ checked_at: "2026-10-03"
234
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
235
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
236
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
237
+ provenance.licence_tiers:
238
+ fill: auto
239
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
240
+ checked_at: "2026-10-03"
241
+ provenance.spdx_top:
242
+ fill: auto
243
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
244
+ checked_at: "2026-10-03"
245
+ risk.contamination:
246
+ fill: auto
247
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
248
+ checked_at: "2026-10-03"
249
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
250
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
251
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
252
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
253
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
255
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
256
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
257
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
260
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
261
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
262
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
265
+ snapshot.rev:
266
+ fill: auto
267
+ basis: "last commit listed in the publish receipt of this source"
268
+ checked_at: "2026-10-03"
269
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
270
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
272
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
273
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
275
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ subset:
280
+ fill: auto
281
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
282
+ checked_at: "2026-10-03"
283
+ unit.pointer_count:
284
+ fill: auto
285
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
286
+ checked_at: "2026-10-03"
287
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
289
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/libreoffice-bugzilla.yaml ADDED
@@ -0,0 +1,323 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: libreoffice-bugzilla
8
+ parent: null
9
+ title: "LibreOffice Bugzilla bugs and attachments"
10
+ description: "All publicly visible bugs and public attachments on LibreOffice Bugzilla (bugs.documentfoundation.org), listed through the Bugzilla REST API in bug number order. One pointer per bug (collection bug) and one per public attachment (collection attachment). Many attachments are the real documents that triggered the bug: OpenDocument, Office Open XML, legacy Microsoft Office, PDF, RTF and CSV files, plus screenshots and patches. For a bug the hash covers its comment list as returned by the API; an attachment is retrieved through the REST API, which returns its content base64 encoded. Product, component, version, status, severity, keywords, attachment media type and an office file category are in source_meta."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Copyright of the contributors; site footer names CC BY-SA 4.0 (patches MPL 2.0)"
20
+ licence_url: "https://bugs.documentfoundation.org/"
21
+ licence_statement: "Bugs, comments and attachments remain the copyright of their submitters. The footer of every LibreOffice Bugzilla page reads: \"Copyright information: Please note that all contributions to The Document Foundation Bugzilla are considered to be released under the Creative Commons Attribution-ShareAlike 4.0 International License, unless otherwise specified. Source code form contributions such as patches are considered to be modifications under the Mozilla Public License v2.0.\" Attachments are often documents made by third parties, which the uploader may not be able to license."
22
+ per_collection: null
23
+ training_ok: "yes"
24
+ redistribution_ok: conditional
25
+ attribution: null
26
+ per_file: null
27
+ licence_tiers: {yes/conditional: 244962}
28
+ spdx_top: {"(none)": 244962}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 244962
34
+ units: {"API response": 244962}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 51118483241
38
+ rows_without_size: 0
39
+ collections:
40
+ distinct: 2
41
+ top: {attachment: 130802, bug: 114160}
42
+ sub_collections: null
43
+ batches: null
44
+ n_ext: 498
45
+ usable_unit: null
46
+ form:
47
+ by_ext:
48
+ json: 114163
49
+ png: 43184
50
+ odt: 15145
51
+ ods: 10094
52
+ jpg: 7497
53
+ docx: 6635
54
+ pdf: 6157
55
+ txt: 5686
56
+ zip: 3534
57
+ odp: 2869
58
+ xlsx: 2796
59
+ doc: 2212
60
+ mp4: 1791
61
+ pptx: 1770
62
+ odg: 1671
63
+ log: 1575
64
+ odb: 1484
65
+ xls: 1202
66
+ svg: 1041
67
+ rtf: 1026
68
+ "(478 other types)": 13430
69
+ bytes_p50: 9610
70
+ bytes_p95: 585749
71
+ bytes_max: 30705906
72
+ size_quantiles:
73
+ max: 30705906
74
+ min: 0
75
+ p25: 2904
76
+ p50: 9610
77
+ p75: 52296
78
+ p90: 212397
79
+ p95: 585749
80
+ p99: 4388380
81
+ p99.9: 21563213
82
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
83
+ by_ext_size:
84
+ - {ext: json, rows: 114163, bytes: 508813164, p50: 2988, p95: 12909, max: 863666}
85
+ - {ext: png, rows: 43184, bytes: 6355966124, p50: 72651, p95: 455289, max: 16232406}
86
+ - {ext: odt, rows: 15145, bytes: 4623541268, p50: 18534, p95: 1048936, max: 29918368}
87
+ - {ext: ods, rows: 10094, bytes: 2335549818, p50: 16361, p95: 590475, max: 30428062}
88
+ - {ext: jpg, rows: 7497, bytes: 1755945538, p50: 98772, p95: 623108, max: 14488498}
89
+ - {ext: docx, rows: 6635, bytes: 1713567609, p50: 22350, p95: 1054736, max: 29732177}
90
+ - {ext: pdf, rows: 6157, bytes: 2111728494, p50: 69710, p95: 1253658, max: 26734920}
91
+ - {ext: txt, rows: 5686, bytes: 673717873, p50: 11161, p95: 239254, max: 28170557}
92
+ - {ext: zip, rows: 3534, bytes: 2809401242, p50: 137008, p95: 3448984, max: 29988229}
93
+ - {ext: odp, rows: 2869, bytes: 2006844328, p50: 28226, p95: 3460740, max: 29634815}
94
+ - {ext: xlsx, rows: 2796, bytes: 768648825, p50: 13234, p95: 908302, max: 26887728}
95
+ - {ext: doc, rows: 2212, bytes: 774003664, p50: 48640, p95: 1615872, max: 24510976}
96
+ - {ext: mp4, rows: 1791, bytes: 8265829090, p50: 1856401, p95: 21195893, max: 30684394}
97
+ - {ext: pptx, rows: 1770, bytes: 1728036023, p50: 58173, p95: 5291453, max: 30693404}
98
+ - {ext: odg, rows: 1671, bytes: 566586261, p50: 19588, p95: 970541, max: 29418029}
99
+ - {ext: log, rows: 1575, bytes: 470755806, p50: 25691, p95: 523791, max: 26409981}
100
+ - {ext: odb, rows: 1484, bytes: 245191070, p50: 12926, p95: 661291, max: 20121683}
101
+ - {ext: xls, rows: 1202, bytes: 485155189, p50: 27648, p95: 1559237, max: 28083712}
102
+ - {ext: svg, rows: 1041, bytes: 380002273, p50: 18969, p95: 2034821, max: 12509090}
103
+ - {ext: rtf, rows: 1026, bytes: 305282010, p50: 42432, p95: 1424549, max: 17081521}
104
+ over_cap:
105
+ cap: 61944279
106
+ rows_over: 0
107
+ bytes_over: 0
108
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
109
+ largest:
110
+ - {pointer_id: "libreoffice-bugzilla:attachment:180207", size_bytes: 30705906}
111
+ - {pointer_id: "libreoffice-bugzilla:attachment:151972", size_bytes: 30693404}
112
+ - {pointer_id: "libreoffice-bugzilla:attachment:176542", size_bytes: 30691918}
113
+ - {pointer_id: "libreoffice-bugzilla:attachment:204380", size_bytes: 30684394}
114
+ - {pointer_id: "libreoffice-bugzilla:attachment:206298", size_bytes: 30639316}
115
+ date_years: null
116
+ media_type:
117
+ application/json: 114161
118
+ application/pdf: 5851
119
+ application/vnd.oasis.opendocument.presentation: 2854
120
+ application/vnd.oasis.opendocument.spreadsheet: 9729
121
+ application/vnd.oasis.opendocument.text: 14959
122
+ application/vnd.openxmlformats-officedocument.spreadsheetml.sheet: 2631
123
+ application/vnd.openxmlformats-officedocument.wordprocessingml.document: 6254
124
+ image/jpeg: 7726
125
+ image/png: 43175
126
+ text/plain: 9343
127
+ facets:
128
+ component:
129
+ "Android Viewer": 533
130
+ BASIC: 1631
131
+ Base: 3601
132
+ Calc: 22675
133
+ Chart: 1548
134
+ Documentation: 2045
135
+ Draw: 3504
136
+ "Formula Editor": 797
137
+ Impress: 8010
138
+ Installation: 1862
139
+ LibreOffice: 14891
140
+ Linguistic: 667
141
+ Localization: 1122
142
+ "Printing and PDF export": 1721
143
+ UI: 7900
144
+ WWW: 558
145
+ Writer: 35614
146
+ "filters and storage": 1406
147
+ framework: 519
148
+ "graphics stack": 1274
149
+ is_obsolete: {"false": 126124, "true": 4678}
150
+ is_patch: {"false": 129440, "true": 1362}
151
+ office_category: {csv: 416, legacy_office: 3924, odf: 32280, ooxml: 11498, other: 75490, pdf: 6166, rtf: 1028}
152
+ product:
153
+ "Document Liberation Project": 239
154
+ "Impress Remote": 205
155
+ LibreOffice: 113022
156
+ "LibreOffice Online": 498
157
+ "QA Tools": 82
158
+ cppunit: 99
159
+ libexttextcat: 15
160
+ resolution:
161
+ DUPLICATE: 22071
162
+ FIXED: 30634
163
+ INSUFFICIENTDATA: 8376
164
+ INVALID: 6803
165
+ MOVED: 415
166
+ NOTABUG: 6261
167
+ NOTOURBUG: 1721
168
+ WONTFIX: 2715
169
+ WORKSFORME: 15743
170
+ severity:
171
+ blocker: 495
172
+ critical: 3305
173
+ enhancement: 14692
174
+ major: 6568
175
+ minor: 7905
176
+ normal: 78875
177
+ trivial: 2320
178
+ status:
179
+ ASSIGNED: 220
180
+ CLOSED: 2869
181
+ NEEDINFO: 405
182
+ NEW: 16079
183
+ REOPENED: 201
184
+ RESOLVED: 83799
185
+ UNCONFIRMED: 2516
186
+ VERIFIED: 8071
187
+ artifact: null
188
+ source_origin: null
189
+ parse_cost: null
190
+ parse_quirks: null
191
+ structure:
192
+ self_describing: null
193
+ integrity_builtin: null
194
+ has_reference_output: null
195
+ invariants: null
196
+ tools: null
197
+ ships:
198
+ problem_statement: {present: null}
199
+ reference_solution: {share: null}
200
+ tests: {present: null}
201
+ env_definition: {present: null}
202
+ known_answer: {present: null}
203
+ generator: {present: null}
204
+ upstream_labels: {present: null}
205
+ class: null
206
+ subject:
207
+ domain_span: null
208
+ shape_span: null
209
+ skill_family_span: null
210
+ ont_domain: null
211
+ tech_vocabulary: null
212
+ risk:
213
+ contamination: none
214
+ overlap_check:
215
+ done: true
216
+ pointers_flagged: 0
217
+ pointers_removed: 0
218
+ by_class: {}
219
+ benches: {}
220
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
221
+ pii: null
222
+ pii_evidence: null
223
+ malware: null
224
+ ship_bytes: conditional
225
+ subset:
226
+ rule: "take all rows; licence, size and version are labels, not filters"
227
+ rows_in_subset: 244962
228
+ last_built: null
229
+ access:
230
+ method: "Bug rows and attachment rows: GET uri; the response is JSON."
231
+ verify: "Bug rows: take bugs.<bug number>.comments (the path is in source_meta.api_body_field; a list of objects with a single text key), serialize it with json.dumps(ensure_ascii=False), encode it as UTF-8 and compare size and sha256; the stored file is that JSON. Attachment rows: take attachments.<attachment number>.data (the path is in source_meta.api_body_field), decode it as base64 (source_meta.api_body_decode) and compare the size of the decoded bytes; the stored file is the decoded attachment. The source publishes no hash, so compute it on download."
232
+ hosts: {"https://bugs.documentfoundation.org": 244962}
233
+ uri_example: "https://bugs.documentfoundation.org/rest/bug/attachment/100000"
234
+ needs_auth: {"false": 244962}
235
+ credential: null
236
+ hash_source: {computed: 114160, computed_on_fetch: 130802}
237
+ range_supported: null
238
+ reachable:
239
+ index_flag: {unknown: 244962}
240
+ index_check: null
241
+ our_sample: null
242
+ drift: null
243
+ sample: null
244
+ evidence:
245
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
246
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
247
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
248
+ class: {fill: unfilled, basis: "ships not judged"}
249
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
250
+ form.by_ext:
251
+ fill: auto
252
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
253
+ checked_at: "2026-10-03"
254
+ form.date_years:
255
+ fill: auto
256
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
257
+ checked_at: "2026-10-03"
258
+ form.over_cap:
259
+ fill: auto
260
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
261
+ checked_at: "2026-10-03"
262
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ form.size_quantiles:
265
+ fill: auto
266
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
267
+ checked_at: "2026-10-03"
268
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
270
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ provenance.licence_tiers:
272
+ fill: auto
273
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
274
+ checked_at: "2026-10-03"
275
+ provenance.spdx_top:
276
+ fill: auto
277
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
278
+ checked_at: "2026-10-03"
279
+ risk.contamination:
280
+ fill: auto
281
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
282
+ checked_at: "2026-10-03"
283
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
287
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
290
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
291
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
292
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
293
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
294
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
295
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
296
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
297
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
298
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
299
+ snapshot.rev:
300
+ fill: auto
301
+ basis: "last commit listed in the publish receipt of this source"
302
+ checked_at: "2026-10-03"
303
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
304
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
305
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
306
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
307
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
308
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
309
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
310
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
311
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
312
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
313
+ subset:
314
+ fill: auto
315
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
316
+ checked_at: "2026-10-03"
317
+ unit.pointer_count:
318
+ fill: auto
319
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
320
+ checked_at: "2026-10-03"
321
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
322
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
323
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/lichess.yaml ADDED
@@ -0,0 +1,497 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: lichess
8
+ parent: null
9
+ title: "Lichess open database"
10
+ description: "Monthly PGN files of standard games and 8 variants, monthly PGN files of broadcast games, the puzzle database (CSV) and the engine evaluation database (JSONL) from database.lichess.org, all zstd-compressed. One pointer per published file."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: e9f6f94b7ffc2827b8376f13aa3b610532d2093b, at: "2026-10-01", rev_candidates: 3}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "CC0 1.0 (broadcast games: CC BY-SA 4.0)"
17
+ licence_url: "https://database.lichess.org"
18
+ licence_statement: "The site states: \"Database exports are released under the Creative Commons CC0 license\". Broadcast games are marked CC BY-SA 4.0."
19
+ per_collection: null
20
+ training_ok: "yes"
21
+ redistribution_ok: "yes"
22
+ attribution: null
23
+ per_file: "Broadcast rows carry license_override with CC BY-SA 4.0."
24
+ licence_tiers: {yes/conditional: 80, yes/yes: 1289}
25
+ spdx_top: {"(none)": 1289, CC-BY-SA-4.0: 80}
26
+ spdx_distinct: 2
27
+ licence_sample_check:
28
+ sample_n: 10
29
+ declared: 0
30
+ unknown_or_missing: 10
31
+ note: "no licence text inside the PGN files; the site states CC0 for database exports"
32
+ unit:
33
+ pointer_level: file
34
+ pointer_count: 1369
35
+ units: {file: 1369}
36
+ state: expanded
37
+ expanded_count: null
38
+ bytes_total: 2632761066822
39
+ rows_without_size: 0
40
+ collections:
41
+ distinct: 12
42
+ top:
43
+ antichess: 141
44
+ atomic: 140
45
+ broadcast: 80
46
+ chess960: 157
47
+ crazyhouse: 128
48
+ horde: 137
49
+ kingOfTheHill: 146
50
+ racingKings: 128
51
+ standard: 164
52
+ threeCheck: 146
53
+ sub_collections: null
54
+ batches: null
55
+ n_ext: 1
56
+ usable_unit: "one game (a PGN tag block plus its move text); a file holds every rated game of that variant in that month"
57
+ form:
58
+ by_ext: {zst: 1369}
59
+ bytes_p50: 21595303
60
+ bytes_p95: 26626823742
61
+ bytes_max: 34864029436
62
+ size_quantiles:
63
+ max: 34864029436
64
+ min: 49180
65
+ p25: 7140199
66
+ p50: 21595303
67
+ p75: 73413914
68
+ p90: 392158685
69
+ p95: 26626823742
70
+ p99: 31614604286
71
+ p99.9: 33724525615
72
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
73
+ by_ext_size:
74
+ - {ext: zst, rows: 1369, bytes: 2632761066822, p50: 21595303, p95: 26626823742, max: 34864029436}
75
+ over_cap:
76
+ cap: 61944279
77
+ rows_over: 390
78
+ bytes_over: 2617051093232
79
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
80
+ largest:
81
+ - {pointer_id: "lichess:standard:lichess_db_standard_rated_2023-03.pgn.zst", size_bytes: 34864029436}
82
+ - {pointer_id: "lichess:standard:lichess_db_standard_rated_2023-05.pgn.zst", size_bytes: 33724525615}
83
+ - {pointer_id: "lichess:standard:lichess_db_standard_rated_2023-01.pgn.zst", size_bytes: 33465852948}
84
+ - {pointer_id: "lichess:standard:lichess_db_standard_rated_2022-01.pgn.zst", size_bytes: 33232239428}
85
+ - {pointer_id: "lichess:standard:lichess_db_standard_rated_2025-01.pgn.zst", size_bytes: 32870953370}
86
+ date_years: null
87
+ media_type: {"(none)": 1369}
88
+ facets:
89
+ etag: {"6aa199fe-12253910": 1, "6aa21979-52475bec9": 1}
90
+ kind: {evals: 1, games: 1367, puzzles: 1}
91
+ last_modified: {"Thu, 10 Sep 2026 02:44:09 GMT": 1, "Wed, 09 Sep 2026 17:40:14 GMT": 1}
92
+ variant:
93
+ antichess: 141
94
+ atomic: 140
95
+ broadcast: 80
96
+ chess960: 157
97
+ crazyhouse: 128
98
+ horde: 137
99
+ kingOfTheHill: 146
100
+ racingKings: 128
101
+ standard: 164
102
+ threeCheck: 146
103
+ artifact: [structured-data]
104
+ source_origin: []
105
+ parse_cost: needs_decompress
106
+ parse_quirks:
107
+ - "each file opens with a zstd skippable frame before the data frame"
108
+ - "a month file can start with games dated the last day of the previous month"
109
+ - "the Date and Round tags are missing in older files, UTCDate is always present"
110
+ - "variant games start from a FEN tag when the start position is not standard"
111
+ - "some games carry [%eval] engine comments and move marks like ?!"
112
+ structure:
113
+ self_describing: true
114
+ integrity_builtin: []
115
+ has_reference_output: false
116
+ invariants:
117
+ - "Result tag matches the game end in the move text"
118
+ - "WhiteRatingDiff and BlackRatingDiff move ratings in opposite directions"
119
+ - "Variant tag matches the collection of the file"
120
+ tools: [zstd, python-chess, pgn-extract, "a chess engine for variants (Fairy-Stockfish)"]
121
+ ships:
122
+ problem_statement: {present: false}
123
+ reference_solution:
124
+ share: {k: 0, "n": 10}
125
+ tests: {present: false, coverage: none, runs: null}
126
+ env_definition: {present: false, kind: none, runs: null}
127
+ known_answer: {present: true}
128
+ generator: {present: false}
129
+ upstream_labels:
130
+ present: true
131
+ fields:
132
+ - Result
133
+ - Termination
134
+ - WhiteElo
135
+ - BlackElo
136
+ - WhiteRatingDiff
137
+ - BlackRatingDiff
138
+ - TimeControl
139
+ - "[%eval] comments"
140
+ class: instance
141
+ subject:
142
+ domain_span: [D10, D10.1]
143
+ shape_span: [S4_data_to_artifact]
144
+ skill_family_span: [skill.data, skill.reasoning]
145
+ ont_domain: [games-puzzles]
146
+ tech_vocabulary: [PGN, FEN, Elo, chess960, atomic, "racing kings", three-check, "king of the hill", zstd]
147
+ risk:
148
+ contamination: none
149
+ overlap_check:
150
+ done: true
151
+ pointers_flagged: 0
152
+ pointers_removed: 0
153
+ by_class: {}
154
+ benches: {}
155
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
156
+ pii: present
157
+ pii_evidence:
158
+ sample_n: 10
159
+ files_with_email: 0
160
+ files_with_user_names: 10
161
+ note: "user names of players in White and Black tags; no emails or addresses"
162
+ malware: null
163
+ ship_bytes: "yes"
164
+ subset:
165
+ rule: "take all rows; licence, size and version are labels, not filters"
166
+ rows_in_subset: 1369
167
+ last_built:
168
+ date: "2026-10-02"
169
+ rule: "older rule that filtered by licence tier"
170
+ rows: 1369
171
+ rows_in_subset: 1369
172
+ excluded_by_licence: 0
173
+ duplicates: 0
174
+ access:
175
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
176
+ verify: "Month files: compare sha256 (published). Puzzle and evaluation files, and any month file without a published hash: compare size, then compute and record sha256."
177
+ hosts: {"https://database.lichess.org": 1369}
178
+ uri_example: "https://database.lichess.org/antichess/lichess_db_antichess_rated_2014-12.pgn.zst"
179
+ needs_auth: {"false": 1369}
180
+ credential: none
181
+ hash_source: {computed_on_fetch: 2, source_published: 1367}
182
+ range_supported: null
183
+ reachable:
184
+ index_flag: {unknown: 1369}
185
+ index_check:
186
+ checked_at: "2026-10-01"
187
+ sample: null
188
+ checked: 1369
189
+ ok: 1369
190
+ by_status: {ok: 1369}
191
+ our_sample: {checked_at: "2026-10-04", n_tried: 10, n_ok: 10, hash_checked: 10}
192
+ drift: "stable: a published month file is not rewritten; a new file appears each month"
193
+ sample:
194
+ method: "strata are file type times the size quartile of this source (q1 to q4 by its own p25, p50, p75; unk for rows without a size); half of the samples go one per stratum to the largest strata, the rest in proportion to stratum size (largest remainder); within a stratum the rows with the smallest sha256(seed:pointer_id) are taken, so the draw is reproducible from the index"
195
+ seed: 20261004
196
+ "n": 10
197
+ n_drawn: 10
198
+ n_read: 10
199
+ unread: []
200
+ rule_n: "10 when one file type holds at least 95 percent of rows, else 30"
201
+ size_quartiles: [7140199, 21595303, 73413914]
202
+ size_filter: "only files of at most 15000000 bytes are candidates"
203
+ strata_candidates: {"zst|q1": 343, "zst|q2": 201}
204
+ strata_total: 2
205
+ strata_taken: {"zst|q1": 343, "zst|q2": 201}
206
+ fetched_at: "2026-10-04"
207
+ replaced: 0
208
+ kept: false
209
+ kept_note: "sample bytes were deleted after reading; refetch by pointer_id and compare sha256"
210
+ files:
211
+ - "no": "#00"
212
+ pointer_id: "lichess:racingKings:lichess_db_racingKings_rated_2018-04.pgn.zst"
213
+ stratum: "zst|q1"
214
+ size_bytes: 1557639
215
+ sha256: "4618ac1756642178955047f252c940630a8281e7b10af9dbe40beb2dc2f6bafe"
216
+ - "no": "#01"
217
+ pointer_id: "lichess:racingKings:lichess_db_racingKings_rated_2017-10.pgn.zst"
218
+ stratum: "zst|q1"
219
+ size_bytes: 1130054
220
+ sha256: "5c968c6451728372ee0778c455802235222f9fd6978da3575bc75f1f507c3def"
221
+ - "no": "#02"
222
+ pointer_id: "lichess:chess960:lichess_db_chess960_rated_2014-11.pgn.zst"
223
+ stratum: "zst|q1"
224
+ size_bytes: 5459465
225
+ sha256: "6270fc9b99ffeac536b64fe8720b6e927d3c6be09d156e47f6284a492f684602"
226
+ - "no": "#03"
227
+ pointer_id: "lichess:threeCheck:lichess_db_threeCheck_rated_2019-11.pgn.zst"
228
+ stratum: "zst|q1"
229
+ size_bytes: 4926653
230
+ sha256: "9860d60fabaf606b6fe2eb7151dcf4faf03422804de7d4a10e378de40ff7008e"
231
+ - "no": "#04"
232
+ pointer_id: "lichess:racingKings:lichess_db_racingKings_rated_2017-12.pgn.zst"
233
+ stratum: "zst|q1"
234
+ size_bytes: 1646100
235
+ sha256: d98d4e10b23ccfc17192ca1eca4afcf7804e0131cd0b13a78cfb3acd3df3b012
236
+ - "no": "#05"
237
+ pointer_id: "lichess:kingOfTheHill:lichess_db_kingOfTheHill_rated_2016-01.pgn.zst"
238
+ stratum: "zst|q1"
239
+ size_bytes: 3869042
240
+ sha256: fd2a9e049caea5362e2d050359b60f3d1e7d0437b86cdeb1a0378128651d3372
241
+ - "no": "#06"
242
+ pointer_id: "lichess:atomic:lichess_db_atomic_rated_2017-12.pgn.zst"
243
+ stratum: "zst|q2"
244
+ size_bytes: 12741328
245
+ sha256: "61b48c0448de4e4d1fe099cc15346164f13c11e594be68cdc95b7477cd23be40"
246
+ - "no": "#07"
247
+ pointer_id: "lichess:kingOfTheHill:lichess_db_kingOfTheHill_rated_2020-07.pgn.zst"
248
+ stratum: "zst|q2"
249
+ size_bytes: 12681411
250
+ sha256: "98d483d51e351891b4f64088503b7ea7f8e4b41e7d8364297f74e9806f1d1825"
251
+ - "no": "#08"
252
+ pointer_id: "lichess:atomic:lichess_db_atomic_rated_2019-09.pgn.zst"
253
+ stratum: "zst|q2"
254
+ size_bytes: 9360724
255
+ sha256: "6ba4cd7404e56ab7f3620ba88a8375557255818b6bc83c112aea71f8b8fff9d3"
256
+ - "no": "#09"
257
+ pointer_id: "lichess:racingKings:lichess_db_racingKings_rated_2025-08.pgn.zst"
258
+ stratum: "zst|q2"
259
+ size_bytes: 7416337
260
+ sha256: "2fcc555e7469fa2457e196adfa7652c770922fcebd9ddc30e716ff09c2a5c678"
261
+ evidence:
262
+ access.credential:
263
+ fill: judgment
264
+ basis: "database.lichess.org serves files without a login"
265
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
266
+ sample_n: 10
267
+ checked_at: "2026-10-04"
268
+ access.drift:
269
+ fill: judgment
270
+ basis: "file names carry the month; the newest sample is 2025-08 (#09)"
271
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
272
+ sample_n: 10
273
+ checked_at: "2026-10-04"
274
+ access.range_supported: {fill: unfilled, basis: "no sample evidence given for this field"}
275
+ class:
276
+ fill: derived
277
+ basis: "rule over ships.tests, env_definition, known_answer, generator and structure; manifest_not_run when tests and a dependency manifest are present but the build was not run"
278
+ form.artifact:
279
+ fill: judgment
280
+ basis: "PGN: tagged game records with move lists, clock and sometimes engine evaluation comments"
281
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
282
+ sample_n: 10
283
+ checked_at: "2026-10-04"
284
+ form.by_ext:
285
+ fill: auto
286
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
287
+ checked_at: "2026-10-03"
288
+ form.date_years:
289
+ fill: auto
290
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
291
+ checked_at: "2026-10-03"
292
+ form.over_cap:
293
+ fill: auto
294
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
295
+ checked_at: "2026-10-03"
296
+ form.parse_cost:
297
+ fill: judgment
298
+ basis: "zstd must be decoded first; the PGN inside is plain text that a chess library parses"
299
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
300
+ sample_n: 10
301
+ checked_at: "2026-10-04"
302
+ form.parse_quirks:
303
+ fill: judgment
304
+ basis: "the first collect run showed the raw heads opening with skippable frame magic 50 2a 4d 18 and the digest decoded nothing until the frame was skipped; UTCDate 2018.03.31 in the 2018-04 file (#00) and 2017.09.30 in the 2017-10 file (#01); Date tag absent in #01, #02, #04, #05, #06; FEN in #00, #01, #02, #04, #09; %eval in #00, #03, #07, #09"
305
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
306
+ sample_n: 10
307
+ checked_at: "2026-10-04"
308
+ form.size_quantiles:
309
+ fill: auto
310
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
311
+ checked_at: "2026-10-03"
312
+ form.source_origin:
313
+ fill: judgment
314
+ basis: "no source code"
315
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
316
+ sample_n: 10
317
+ checked_at: "2026-10-04"
318
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
319
+ provenance.licence_sample_check:
320
+ fill: judgment
321
+ basis: "no licence tag in the heads"
322
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
323
+ sample_n: 10
324
+ checked_at: "2026-10-04"
325
+ provenance.licence_tiers:
326
+ fill: auto
327
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
328
+ checked_at: "2026-10-03"
329
+ provenance.spdx_top:
330
+ fill: auto
331
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
332
+ checked_at: "2026-10-03"
333
+ risk.contamination:
334
+ fill: auto
335
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
336
+ checked_at: "2026-10-03"
337
+ risk.malware: {fill: unfilled, basis: "no sample evidence given for this field"}
338
+ risk.pii:
339
+ fill: judgment
340
+ basis: "ruling after independent re-judge: every game names both players by user name and some handles read as real personal names (#00 'ignacykowalski'); the field definition counts a public name of a person as present; re-checked under the WP-T-44c definition (a user name on a platform counts as present)"
341
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
342
+ sample_n: 10
343
+ checked_at: "2026-10-04"
344
+ risk.pii_evidence:
345
+ fill: judgment
346
+ basis: "email regex and the White and Black tags"
347
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
348
+ sample_n: 10
349
+ checked_at: "2026-10-04"
350
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
351
+ ships.env_definition.kind:
352
+ fill: judgment
353
+ basis: "no environment definition"
354
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
355
+ sample_n: 10
356
+ checked_at: "2026-10-04"
357
+ ships.env_definition.present:
358
+ fill: judgment
359
+ basis: "no environment definition"
360
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
361
+ sample_n: 10
362
+ checked_at: "2026-10-04"
363
+ ships.env_definition.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
364
+ ships.generator.present:
365
+ fill: judgment
366
+ basis: "no generator"
367
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
368
+ sample_n: 10
369
+ checked_at: "2026-10-04"
370
+ ships.known_answer.present:
371
+ fill: judgment
372
+ basis: "ruling after independent re-judge: every game carries a Result tag (1-0, 0-1, 1/2-1/2; 66 games across the 10 heads) that a checker can compare with the outcome of replaying the moves"
373
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
374
+ sample_n: 10
375
+ checked_at: "2026-10-04"
376
+ ships.problem_statement.present:
377
+ fill: judgment
378
+ basis: "game records, not tasks"
379
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
380
+ sample_n: 10
381
+ checked_at: "2026-10-04"
382
+ ships.reference_solution.share:
383
+ fill: judgment
384
+ basis: "0 of 10 sample files read: no sample file carries a fix, patch, solved answer or worked solution"
385
+ files: []
386
+ sample_n: 10
387
+ checked_at: "2026-10-04"
388
+ ships.tests.coverage:
389
+ fill: judgment
390
+ basis: "no tests"
391
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
392
+ sample_n: 10
393
+ checked_at: "2026-10-04"
394
+ ships.tests.present:
395
+ fill: judgment
396
+ basis: "no tests"
397
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
398
+ sample_n: 10
399
+ checked_at: "2026-10-04"
400
+ ships.tests.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
401
+ ships.upstream_labels.fields:
402
+ fill: judgment
403
+ basis: "tags seen in all 10 heads; %eval in #00, #03, #07, #09"
404
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
405
+ sample_n: 10
406
+ checked_at: "2026-10-04"
407
+ ships.upstream_labels.present:
408
+ fill: judgment
409
+ basis: "every game carries its result, termination reason and player ratings; some carry engine evaluations"
410
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
411
+ sample_n: 10
412
+ checked_at: "2026-10-04"
413
+ snapshot.rev:
414
+ fill: auto
415
+ basis: "last commit listed in the publish receipt of this source"
416
+ checked_at: "2026-10-03"
417
+ structure.has_reference_output:
418
+ fill: judgment
419
+ basis: "no expected output file"
420
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
421
+ sample_n: 10
422
+ checked_at: "2026-10-04"
423
+ structure.integrity_builtin:
424
+ fill: judgment
425
+ basis: "no checksum inside the files (sha256 is published by the site)"
426
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
427
+ sample_n: 10
428
+ checked_at: "2026-10-04"
429
+ structure.invariants:
430
+ fill: judgment
431
+ basis: "checked by eye on the heads of all 10"
432
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
433
+ sample_n: 10
434
+ checked_at: "2026-10-04"
435
+ structure.self_describing:
436
+ fill: judgment
437
+ basis: "every game names its fields in PGN tags"
438
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
439
+ sample_n: 10
440
+ checked_at: "2026-10-04"
441
+ structure.tools:
442
+ fill: judgment
443
+ basis: "PGN parsing and variant rules"
444
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
445
+ sample_n: 10
446
+ checked_at: "2026-10-04"
447
+ subject.domain_span:
448
+ fill: judgment
449
+ basis: "chess variant games: racing kings, chess960, three-check, king of the hill, atomic"
450
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
451
+ sample_n: 10
452
+ checked_at: "2026-10-04"
453
+ subject.ont_domain:
454
+ fill: judgment
455
+ basis: "all samples are chess games"
456
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
457
+ sample_n: 10
458
+ checked_at: "2026-10-04"
459
+ subject.shape_span:
460
+ fill: judgment
461
+ basis: "parse game collections into statistics, positions or filtered sets"
462
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
463
+ sample_n: 10
464
+ checked_at: "2026-10-04"
465
+ subject.skill_family_span:
466
+ fill: judgment
467
+ basis: "PGN processing; game rules and position reasoning"
468
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
469
+ sample_n: 10
470
+ checked_at: "2026-10-04"
471
+ subject.tech_vocabulary:
472
+ fill: judgment
473
+ basis: "terms in the tags and file names"
474
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
475
+ sample_n: 10
476
+ checked_at: "2026-10-04"
477
+ subset:
478
+ fill: auto
479
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
480
+ checked_at: "2026-10-03"
481
+ unit.pointer_count:
482
+ fill: auto
483
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
484
+ checked_at: "2026-10-03"
485
+ unit.pointer_level:
486
+ fill: judgment
487
+ basis: "each pointer is one monthly zstd-compressed PGN file of one variant"
488
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
489
+ sample_n: 10
490
+ checked_at: "2026-10-04"
491
+ unit.state: {fill: derived, basis: "packed when pointer_level is archive, else expanded"}
492
+ unit.usable_unit:
493
+ fill: judgment
494
+ basis: "4 to 12 games in each 6000-character head; tags Event, Site, White, Black, Result, Elo and move text in all 10"
495
+ files: ["#00", "#01", "#02", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
496
+ sample_n: 10
497
+ checked_at: "2026-10-04"
cards/linux-kernel-mail-archives.yaml ADDED
@@ -0,0 +1,242 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: linux-kernel-mail-archives
8
+ parent: null
9
+ title: "Linux kernel mailing list archives (lore.kernel.org)"
10
+ description: "Public-inbox git repositories behind lore.kernel.org, listed from the grokmirror manifest that the official mirror source erol.kernel.org publishes for mirroring. One pointer per list per git epoch."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: aa9b603fcfb27e1ac0d0e3fbd522b38e9bfbac84, at: "2026-10-03", rev_candidates: 2}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "No content license stated"
17
+ licence_url: "https://www.kernel.org/lore.html"
18
+ licence_statement: "Messages remain the copyright of their senders; the archive states no content license. The robots.txt of erol.kernel.org disallows all automated crawlers and names AI training crawlers. Not for training."
19
+ per_collection: null
20
+ training_ok: "no"
21
+ redistribution_ok: unknown
22
+ attribution: null
23
+ per_file: null
24
+ licence_tiers: {no/unknown: 405}
25
+ spdx_top: {"(none)": 405}
26
+ spdx_distinct: 1
27
+ licence_sample_check: null
28
+ unit:
29
+ pointer_level: null
30
+ pointer_count: 405
31
+ units: {file: 405}
32
+ state: null
33
+ expanded_count: null
34
+ bytes_total: 0
35
+ rows_without_size: 405
36
+ collections:
37
+ distinct: 356
38
+ top:
39
+ dpdk-dev: 2
40
+ dri-devel: 3
41
+ git: 2
42
+ linux-arm-kernel: 4
43
+ linux-devicetree: 4
44
+ linux-mm: 3
45
+ lkml: 21
46
+ netdev: 4
47
+ qemu-devel: 4
48
+ stable: 3
49
+ sub_collections: null
50
+ batches: null
51
+ n_ext: 1
52
+ usable_unit: null
53
+ form:
54
+ by_ext: {git: 405}
55
+ bytes_p50: null
56
+ bytes_p95: null
57
+ bytes_max: null
58
+ size_quantiles: null
59
+ by_ext_size:
60
+ - {ext: git, rows: 405, bytes: 0, p50: null, p95: null, max: null}
61
+ over_cap:
62
+ cap: 61944279
63
+ rows_over: 0
64
+ bytes_over: 0
65
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
66
+ largest: []
67
+ date_years: null
68
+ media_type: {"(none)": 405}
69
+ facets:
70
+ forkgroup:
71
+ "335b8361-03e0-4711-9954-197385f3cfa0": 2
72
+ "3bd6457f-8550-4123-9899-83d298927204": 2
73
+ "4d6a3d2a-bea6-4439-809b-6b061464465c": 5
74
+ "772dc216-c869-47b5-8015-64509ad6c8c2": 2
75
+ "8117cadd-d887-417b-94b8-ca6774433bb8": 2
76
+ bb2a1b1f-acbc-421b-b52a-8098f9373164: 2
77
+ bb751ba5-a8f5-4433-90a2-dcb623b87101: 2
78
+ c12404af-2ca3-4897-aeef-d14bdd275862: 3
79
+ d69667cc-97c1-461d-ab48-c2f0405ce195: 1
80
+ e0758afe-c762-4009-a455-888b89f5b03a: 5
81
+ f4eefcb8-a4da-4ab3-b889-673e8ae97118: 2
82
+ f8644b5f-2e97-4316-b9fc-56e105e64ccd: 2
83
+ reference:
84
+ "/dri-devel/git/1.git": 1
85
+ "/intel-gfx/git/1.git": 1
86
+ "/ksummit/git/0.git": 1
87
+ "/linux-arm-kernel/git/0.git": 1
88
+ "/linux-devicetree/git/0.git": 1
89
+ "/lkml/git/2.git": 4
90
+ "/lkml/git/7.git": 4
91
+ "/netdev/git/0.git": 1
92
+ "/nvdimm/git/0.git": 1
93
+ "/qemu-devel/git/1.git": 2
94
+ "/xen-devel/git/1.git": 1
95
+ artifact: null
96
+ source_origin: null
97
+ parse_cost: null
98
+ parse_quirks: null
99
+ structure:
100
+ self_describing: null
101
+ integrity_builtin: null
102
+ has_reference_output: null
103
+ invariants: null
104
+ tools: null
105
+ ships:
106
+ problem_statement: {present: null}
107
+ reference_solution: {share: null}
108
+ tests: {present: null}
109
+ env_definition: {present: null}
110
+ known_answer: {present: null}
111
+ generator: {present: null}
112
+ upstream_labels: {present: null}
113
+ class: null
114
+ subject:
115
+ domain_span: null
116
+ shape_span: null
117
+ skill_family_span: null
118
+ ont_domain: null
119
+ tech_vocabulary: null
120
+ risk:
121
+ contamination: none
122
+ overlap_check:
123
+ done: true
124
+ pointers_flagged: 0
125
+ pointers_removed: 0
126
+ by_class: {}
127
+ benches: {}
128
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
129
+ pii: null
130
+ pii_evidence: null
131
+ malware: null
132
+ ship_bytes: unknown
133
+ subset:
134
+ rule: "take all rows; licence, size and version are labels, not filters"
135
+ rows_in_subset: 405
136
+ last_built:
137
+ date: "2026-10-02"
138
+ rule: "older rule that filtered by licence tier"
139
+ rows: 405
140
+ rows_in_subset: 0
141
+ excluded_by_licence: 405
142
+ duplicates: 0
143
+ access:
144
+ method: "Each uri is a git repository: drop the git+ prefix and clone it, for example git clone --mirror <repository> <directory>."
145
+ verify: "The sha1 of the git show-ref output (one line per reference, commit and name, sorted by name, without HEAD and peeled ^{} lines) must equal source_meta.fingerprint. The source publishes no size; size_bytes is null and the size is measured on retrieval."
146
+ hosts: {"git+https://erol.kernel.org": 405}
147
+ uri_example: "git+https://erol.kernel.org/accel-config/git/0.git"
148
+ needs_auth: {"false": 405}
149
+ credential: null
150
+ hash_source: {computed_on_fetch: 405}
151
+ range_supported: null
152
+ reachable:
153
+ index_flag: {unknown: 405}
154
+ index_check:
155
+ checked_at: "2026-10-01"
156
+ sample: 30
157
+ checked: 405
158
+ ok: 0
159
+ by_status: {error: 405}
160
+ our_sample: null
161
+ drift: null
162
+ sample: null
163
+ evidence:
164
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
165
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
166
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
167
+ class: {fill: unfilled, basis: "ships not judged"}
168
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
169
+ form.by_ext:
170
+ fill: auto
171
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
172
+ checked_at: "2026-10-03"
173
+ form.date_years:
174
+ fill: auto
175
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
176
+ checked_at: "2026-10-03"
177
+ form.over_cap:
178
+ fill: auto
179
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
180
+ checked_at: "2026-10-03"
181
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
182
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
183
+ form.size_quantiles:
184
+ fill: auto
185
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
186
+ checked_at: "2026-10-03"
187
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
188
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
189
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
190
+ provenance.licence_tiers:
191
+ fill: auto
192
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
193
+ checked_at: "2026-10-03"
194
+ provenance.spdx_top:
195
+ fill: auto
196
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
197
+ checked_at: "2026-10-03"
198
+ risk.contamination:
199
+ fill: auto
200
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
201
+ checked_at: "2026-10-03"
202
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
203
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
204
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
205
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
206
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
207
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
208
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
209
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
210
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
211
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
212
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
213
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
214
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
215
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
216
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
217
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
218
+ snapshot.rev:
219
+ fill: auto
220
+ basis: "last commit listed in the publish receipt of this source"
221
+ checked_at: "2026-10-03"
222
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
224
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
225
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
226
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
227
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
228
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
229
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
230
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
231
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
232
+ subset:
233
+ fill: auto
234
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
235
+ checked_at: "2026-10-03"
236
+ unit.pointer_count:
237
+ fill: auto
238
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
239
+ checked_at: "2026-10-03"
240
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
241
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
242
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/loghub.yaml ADDED
@@ -0,0 +1,460 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: loghub
8
+ parent: null
9
+ title: "Loghub: A Large Collection of System Log Datasets for AI-driven Log Analytics (Zenodo record 8196385)"
10
+ description: "Files of Zenodo record 8196385: Loghub: A Large Collection of System Log Datasets for AI-driven Log Analytics (DOI 10.5281/zenodo.8196385). One pointer per file."
11
+ catalog: terminal-data-pointers
12
+ snapshot: {rev: "67f8a9c375c3dcc39b5559f1e0dde2ca5efc787f", at: "2026-10-03", rev_candidates: 3}
13
+ audit: {status: draft, reviewed_by: null}
14
+ provenance:
15
+ url: null
16
+ licence: "Loghub license: research or academic use (Zenodo metadata: CC BY 4.0)"
17
+ licence_url: "https://zenodo.org/records/8196385"
18
+ licence_statement: "The Zenodo record metadata gives CC BY 4.0, while the record description and the LICENSE file of the Loghub repository (https://github.com/logpai/loghub) state: \"The datasets are freely available for research or academic work, subject to the following condition: For any usage or distribution of the loghub datasets, please refer to the loghub repository URL (https://github.com/logpai/loghub) and cite the loghub paper (Loghub: A Large Collection of System Log Datasets for AI-driven Log Analytics) where applicable.\" Training use and redistribution are therefore marked conditional."
19
+ per_collection: null
20
+ training_ok: conditional
21
+ redistribution_ok: conditional
22
+ attribution: "Refer to https://github.com/logpai/loghub, cite the Loghub paper (ISSRE 2023) and include the license notice in all copies."
23
+ per_file: null
24
+ licence_tiers: {conditional/conditional: 19}
25
+ spdx_top: {"(none)": 19}
26
+ spdx_distinct: 1
27
+ licence_sample_check:
28
+ sample_n: 9
29
+ declared: 0
30
+ unknown_or_missing: 9
31
+ note: "no licence file inside the archives; the source-level terms (research or academic use, Zenodo metadata CC BY 4.0) apply"
32
+ unit:
33
+ pointer_level: archive
34
+ pointer_count: 19
35
+ units: {file: 19}
36
+ state: packed
37
+ expanded_count: null
38
+ bytes_total: 5996069686
39
+ rows_without_size: 0
40
+ collections: null
41
+ sub_collections: null
42
+ batches: null
43
+ n_ext: 2
44
+ usable_unit: "one system's log file (or, for Hadoop, one application directory of container logs) plus any label file"
45
+ form:
46
+ by_ext: {gz: 11, zip: 8}
47
+ bytes_p50: 5394691
48
+ bytes_p95: 2016100298
49
+ bytes_max: 2016100298
50
+ size_quantiles:
51
+ max: 2016100298
52
+ min: 172346
53
+ p25: 1489372
54
+ p50: 5394691
55
+ p75: 444986139
56
+ p90: 1670098945
57
+ p95: 2016100298
58
+ p99: 2016100298
59
+ p99.9: 2016100298
60
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
61
+ by_ext_size:
62
+ - {ext: gz, rows: 11, bytes: 3884578534, p50: 2296618, p95: 2016100298, max: 2016100298}
63
+ - {ext: zip, rows: 8, bytes: 2111491152, p50: 57489019, p95: 823700713, max: 823700713}
64
+ over_cap:
65
+ cap: 61944279
66
+ rows_over: 7
67
+ bytes_over: 5892425795
68
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
69
+ largest:
70
+ - {pointer_id: "loghub:zenodo-8196385:Thunderbird.tar.gz", size_bytes: 2016100298}
71
+ - {pointer_id: "loghub:zenodo-8196385:Windows.tar.gz", size_bytes: 1670098945}
72
+ - {pointer_id: "loghub:zenodo-8196385:HDFS_v2.zip", size_bytes: 823700713}
73
+ - {pointer_id: "loghub:zenodo-8196385:HDFS_v3_TraceBench.zip", size_bytes: 567419398}
74
+ - {pointer_id: "loghub:zenodo-8196385:Android_v2.zip", size_bytes: 444986139}
75
+ date_years: null
76
+ media_type: {"(none)": 19}
77
+ facets: null
78
+ artifact: [logs-traces]
79
+ source_origin: []
80
+ parse_cost: needs_decompress
81
+ parse_quirks:
82
+ - "each system uses its own line format (syslog, log4j, pipe-separated, bracketed timestamps)"
83
+ - "syslog-style lines omit the year"
84
+ - "one archive can unpack to 70 MB of text"
85
+ - "Hadoop logs are split across hundreds of container files"
86
+ structure:
87
+ self_describing: false
88
+ integrity_builtin: []
89
+ has_reference_output: true
90
+ invariants:
91
+ - "timestamps non-decreasing within a log"
92
+ - "OpenStack anomaly instance ids appear in openstack_abnormal.log"
93
+ tools: [tar, unzip, grep, awk, "log parsers (Drain)"]
94
+ ships:
95
+ problem_statement: {present: false}
96
+ reference_solution:
97
+ share: {k: 0, "n": 9}
98
+ tests: {present: false, coverage: none, runs: null}
99
+ env_definition: {present: false, kind: none, runs: null}
100
+ known_answer: {present: true}
101
+ generator: {present: false}
102
+ upstream_labels:
103
+ present: true
104
+ fields:
105
+ - "anomaly_labels.txt (OpenStack instance ids)"
106
+ - "abnormal_label.txt (Hadoop application ids by failure type)"
107
+ class: instance
108
+ subject:
109
+ domain_span: [D3.2, D2.6, D8.1]
110
+ shape_span: [S4_data_to_artifact, S3_forensic_recovery]
111
+ skill_family_span: [skill.data, skill.systems, skill.security]
112
+ ont_domain: [software-systems]
113
+ tech_vocabulary: [syslog, kernel, sshd, pam_unix, ZooKeeper, "OpenStack nova", "Hadoop YARN container", "HPC node"]
114
+ risk:
115
+ contamination: none
116
+ overlap_check:
117
+ done: true
118
+ pointers_flagged: 0
119
+ pointers_removed: 0
120
+ by_class: {}
121
+ benches: {}
122
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
123
+ pii: suspected
124
+ pii_evidence:
125
+ sample_n: 9
126
+ files_with_email: 0
127
+ files_with_public_ip_addresses: 1
128
+ note: "public client IPs only in #06; #07 is per-person activity data; #03 and #09 show only 0.0.0.0 and internal node addresses"
129
+ malware: null
130
+ ship_bytes: conditional
131
+ subset:
132
+ rule: "take all rows; licence, size and version are labels, not filters"
133
+ rows_in_subset: 19
134
+ last_built:
135
+ date: "2026-10-02"
136
+ rule: "older rule that filtered by licence tier"
137
+ rows: 19
138
+ rows_in_subset: 19
139
+ excluded_by_licence: 0
140
+ duplicates: 0
141
+ access:
142
+ method: "Download each uri, or an address from alt_uris. When a row has range, request only those bytes with an HTTP Range header."
143
+ verify: "Compare size_bytes and md5 (from the Zenodo API)."
144
+ hosts: {"https://zenodo.org": 19}
145
+ uri_example: "https://zenodo.org/records/8196385/files/Android_v1.zip"
146
+ needs_auth: {"false": 19}
147
+ credential: none
148
+ hash_source: {source_published: 19}
149
+ range_supported: null
150
+ reachable:
151
+ index_flag: {"yes": 19}
152
+ index_check:
153
+ checked_at: "2026-10-01"
154
+ sample: null
155
+ checked: 19
156
+ ok: 19
157
+ by_status: {ok: 19}
158
+ our_sample: {checked_at: "2026-10-04", n_tried: 10, n_ok: 9, hash_checked: 9}
159
+ drift: "stable: a published Zenodo record version does not change; new versions get new record ids"
160
+ sample:
161
+ method: "strata are file type times the size quartile of this source (q1 to q4 by its own p25, p50, p75; unk for rows without a size); half of the samples go one per stratum to the largest strata, the rest in proportion to stratum size (largest remainder); within a stratum the rows with the smallest sha256(seed:pointer_id) are taken, so the draw is reproducible from the index"
162
+ seed: 20261004
163
+ "n": 30
164
+ n_drawn: 10
165
+ n_read: 9
166
+ unread:
167
+ - {"no": "#02", stratum: "gz|q1", status: no_fetchable_row}
168
+ rule_n: "10 when one file type holds at least 95 percent of rows, else 30"
169
+ size_quartiles: [1489372, 5394691, 444986139]
170
+ size_filter: "only files of at most 15000000 bytes are candidates"
171
+ strata_candidates: {"gz|q1": 5, "gz|q2": 3, "zip|q2": 2}
172
+ strata_total: 3
173
+ strata_taken: {"gz|q1": 5, "gz|q2": 3, "zip|q2": 2}
174
+ fetched_at: "2026-10-04"
175
+ replaced: 0
176
+ kept: false
177
+ kept_note: "sample bytes were deleted after reading; refetch by pointer_id and compare sha256"
178
+ files:
179
+ - "no": "#00"
180
+ pointer_id: "loghub:zenodo-8196385:Mac.tar.gz"
181
+ stratum: "gz|q1"
182
+ size_bytes: 1489372
183
+ sha256: a3ebaae341af38f397924cc4f00f2e0e95ff9db4e97abd0dd23986eda239a5a3
184
+ - "no": "#01"
185
+ pointer_id: "loghub:zenodo-8196385:Linux.tar.gz"
186
+ stratum: "gz|q1"
187
+ size_bytes: 232039
188
+ sha256: "7e1f820d8d45ae086032e515d9ce8d079a102a9b37dcff5b41a8e60b1f857820"
189
+ - "no": "#03"
190
+ pointer_id: "loghub:zenodo-8196385:Zookeeper.tar.gz"
191
+ stratum: "gz|q1"
192
+ size_bytes: 459911
193
+ sha256: ea350b0d3ff22bca4764ae4a1521bd1b79c448db4ee99aac02fd82b9a0995f19
194
+ - "no": "#04"
195
+ pointer_id: "loghub:zenodo-8196385:Proxifier.tar.gz"
196
+ stratum: "gz|q1"
197
+ size_bytes: 172346
198
+ sha256: b56d93eb87188e1284b5b35e92cb6f4151aec22975d82c5387d8e81a52d2a96d
199
+ - "no": "#05"
200
+ pointer_id: "loghub:zenodo-8196385:OpenStack.tar.gz"
201
+ stratum: "gz|q2"
202
+ size_bytes: 5394691
203
+ sha256: "87c98c5ed03262e05cdb7a6f3717033df76d88fda0f7d2db23bd9fa4200f1879"
204
+ - "no": "#06"
205
+ pointer_id: "loghub:zenodo-8196385:SSH.tar.gz"
206
+ stratum: "gz|q2"
207
+ size_bytes: 4596666
208
+ sha256: "296610a35773f58b7c295e5407fc5ccdf48beb3c544fc14a6309addf904a2712"
209
+ - "no": "#07"
210
+ pointer_id: "loghub:zenodo-8196385:HealthApp.tar.gz"
211
+ stratum: "gz|q2"
212
+ size_bytes: 2296618
213
+ sha256: "157f5ae299b2cc3ece65d61541ade6f5f4282d428bb0cddb50f6b466d3ee6ffd"
214
+ - "no": "#08"
215
+ pointer_id: "loghub:zenodo-8196385:Hadoop.zip"
216
+ stratum: "zip|q2"
217
+ size_bytes: 3416419
218
+ sha256: "79e63c6521e90ae164754a29d13e04e0c6c5e782490a7fe1e2b124f4f5737ced"
219
+ - "no": "#09"
220
+ pointer_id: "loghub:zenodo-8196385:HPC.zip"
221
+ stratum: "zip|q2"
222
+ size_bytes: 2975898
223
+ sha256: "3df2d31235c441083d8bee41f15bebd9ca5997f1a2ba9e6a967118865dfb1811"
224
+ evidence:
225
+ access.credential:
226
+ fill: judgment
227
+ basis: "Zenodo files download without a login"
228
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
229
+ sample_n: 9
230
+ checked_at: "2026-10-04"
231
+ access.drift:
232
+ fill: judgment
233
+ basis: "gzip timestamps from 2018 (#00 to #07) inside record 8196385"
234
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
235
+ sample_n: 9
236
+ checked_at: "2026-10-04"
237
+ access.range_supported: {fill: unfilled, basis: "no sample evidence given for this field"}
238
+ class:
239
+ fill: derived
240
+ basis: "rule over ships.tests, env_definition, known_answer, generator and structure; manifest_not_run when tests and a dependency manifest are present but the build was not run"
241
+ form.artifact:
242
+ fill: judgment
243
+ basis: "every sample is a raw system or application log"
244
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
245
+ sample_n: 9
246
+ checked_at: "2026-10-04"
247
+ form.by_ext:
248
+ fill: auto
249
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
250
+ checked_at: "2026-10-03"
251
+ form.date_years:
252
+ fill: auto
253
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
254
+ checked_at: "2026-10-03"
255
+ form.over_cap:
256
+ fill: auto
257
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
258
+ checked_at: "2026-10-03"
259
+ form.parse_cost:
260
+ fill: judgment
261
+ basis: "archives must be unpacked; the logs inside are plain text lines"
262
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
263
+ sample_n: 9
264
+ checked_at: "2026-10-04"
265
+ form.parse_quirks:
266
+ fill: judgment
267
+ basis: "formats differ across #00, #03, #04, #07, #09; Mac and Linux lines start with 'Jul 1 09:00:55' (#00, #01); SSH.log is 73417506 bytes (#06); 1035 members in #08"
268
+ files: ["#00", "#01", "#03", "#04", "#06", "#07", "#08", "#09"]
269
+ sample_n: 9
270
+ checked_at: "2026-10-04"
271
+ form.size_quantiles:
272
+ fill: auto
273
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
274
+ checked_at: "2026-10-03"
275
+ form.source_origin:
276
+ fill: judgment
277
+ basis: "no source code"
278
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
279
+ sample_n: 9
280
+ checked_at: "2026-10-04"
281
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
282
+ provenance.licence_sample_check:
283
+ fill: judgment
284
+ basis: "member lists show no LICENSE file"
285
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
286
+ sample_n: 9
287
+ checked_at: "2026-10-04"
288
+ provenance.licence_tiers:
289
+ fill: auto
290
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
291
+ checked_at: "2026-10-03"
292
+ provenance.spdx_top:
293
+ fill: auto
294
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
295
+ checked_at: "2026-10-03"
296
+ risk.contamination:
297
+ fill: auto
298
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
299
+ checked_at: "2026-10-03"
300
+ risk.malware: {fill: unfilled, basis: "no sample evidence given for this field"}
301
+ risk.pii:
302
+ fill: judgment
303
+ basis: "SSH.log carries client IP addresses and the account names tried in login attempts (#06); the phone health app log records one device owner's step counts and calories (#07); Mac and Proxifier logs name hosts and the browsing proxy of one university (#00, #04); no names or emails seen; re-checked under the WP-T-44c definition (a user name on a platform counts as present): the user names in #00 and #06 are names tried by failed SSH logins (webmaster, test9), not accounts of people; stays suspected for the per-person activity data in #07"
304
+ files: ["#00", "#04", "#06", "#07"]
305
+ sample_n: 9
306
+ checked_at: "2026-10-04"
307
+ risk.pii_evidence:
308
+ fill: judgment
309
+ basis: "email and IPv4 regex over the heads"
310
+ files: ["#03", "#06", "#07", "#09"]
311
+ sample_n: 9
312
+ checked_at: "2026-10-04"
313
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
314
+ ships.env_definition.kind:
315
+ fill: judgment
316
+ basis: "no environment definition"
317
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
318
+ sample_n: 9
319
+ checked_at: "2026-10-04"
320
+ ships.env_definition.present:
321
+ fill: judgment
322
+ basis: "no environment definition"
323
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
324
+ sample_n: 9
325
+ checked_at: "2026-10-04"
326
+ ships.env_definition.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
327
+ ships.generator.present:
328
+ fill: judgment
329
+ basis: "no generator"
330
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
331
+ sample_n: 9
332
+ checked_at: "2026-10-04"
333
+ ships.known_answer.present:
334
+ fill: judgment
335
+ basis: "OpenStack lists the VM instances with injected anomalies (#05); Hadoop lists which application runs were normal and which had injected failures such as machine down (#08); the other systems ship no labels"
336
+ files: ["#05", "#08"]
337
+ sample_n: 9
338
+ checked_at: "2026-10-04"
339
+ ships.problem_statement.present:
340
+ fill: judgment
341
+ basis: "logs, not tasks"
342
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
343
+ sample_n: 9
344
+ checked_at: "2026-10-04"
345
+ ships.reference_solution.share:
346
+ fill: judgment
347
+ basis: "0 of 9 sample files read: no sample file carries a fix, patch, solved answer or worked solution"
348
+ files: []
349
+ sample_n: 9
350
+ checked_at: "2026-10-04"
351
+ ships.tests.coverage:
352
+ fill: judgment
353
+ basis: "no tests"
354
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
355
+ sample_n: 9
356
+ checked_at: "2026-10-04"
357
+ ships.tests.present:
358
+ fill: judgment
359
+ basis: "no tests"
360
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
361
+ sample_n: 9
362
+ checked_at: "2026-10-04"
363
+ ships.tests.runs: {fill: unfilled, basis: "no sample evidence given for this field"}
364
+ ships.upstream_labels.fields:
365
+ fill: judgment
366
+ basis: "member names and heads in #05 and #08"
367
+ files: ["#05", "#08"]
368
+ sample_n: 9
369
+ checked_at: "2026-10-04"
370
+ ships.upstream_labels.present:
371
+ fill: judgment
372
+ basis: "the anomaly label files of OpenStack and Hadoop"
373
+ files: ["#05", "#08"]
374
+ sample_n: 9
375
+ checked_at: "2026-10-04"
376
+ snapshot.rev:
377
+ fill: auto
378
+ basis: "last commit listed in the publish receipt of this source"
379
+ checked_at: "2026-10-03"
380
+ structure.has_reference_output:
381
+ fill: judgment
382
+ basis: "ruling after independent re-judge: OpenStack anomaly_labels.txt (#05) and Hadoop abnormal_label.txt (#08) are expected outputs a checker can compare against; only some datasets carry them"
383
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
384
+ sample_n: 9
385
+ checked_at: "2026-10-04"
386
+ structure.integrity_builtin:
387
+ fill: judgment
388
+ basis: "no checksum inside the archives (md5 is published by Zenodo outside them)"
389
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
390
+ sample_n: 9
391
+ checked_at: "2026-10-04"
392
+ structure.invariants:
393
+ fill: judgment
394
+ basis: "timestamps rise in every head shown; the label file says the listed VMs are in the abnormal log (#05)"
395
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
396
+ sample_n: 9
397
+ checked_at: "2026-10-04"
398
+ structure.self_describing:
399
+ fill: judgment
400
+ basis: "no header or schema; the line format must be inferred per system (README in #09 only cites papers)"
401
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
402
+ sample_n: 9
403
+ checked_at: "2026-10-04"
404
+ structure.tools:
405
+ fill: judgment
406
+ basis: "unpack then line-oriented text processing"
407
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
408
+ sample_n: 9
409
+ checked_at: "2026-10-04"
410
+ subject.domain_span:
411
+ fill: judgment
412
+ basis: "log and event analysis on every sample; failure diagnosis on the labelled OpenStack and Hadoop runs (#05, #08) and the HPC errors (#09); SSH brute-force attempts suit incident analysis (#06); trimmed to at most three main values (WP-T-44c): kept the values most samples show, dropped a parent code when its subcode is kept"
413
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
414
+ sample_n: 9
415
+ checked_at: "2026-10-04"
416
+ subject.ont_domain:
417
+ fill: judgment
418
+ basis: "operating system, distributed system and application logs"
419
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
420
+ sample_n: 9
421
+ checked_at: "2026-10-04"
422
+ subject.shape_span:
423
+ fill: judgment
424
+ basis: "turn raw logs into counts, templates or anomaly reports; find the failing run or the intruder"
425
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
426
+ sample_n: 9
427
+ checked_at: "2026-10-04"
428
+ subject.skill_family_span:
429
+ fill: judgment
430
+ basis: "text processing of logs, system operations knowledge, intrusion patterns in SSH.log"
431
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
432
+ sample_n: 9
433
+ checked_at: "2026-10-04"
434
+ subject.tech_vocabulary:
435
+ fill: judgment
436
+ basis: "terms in the log heads"
437
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
438
+ sample_n: 9
439
+ checked_at: "2026-10-04"
440
+ subset:
441
+ fill: auto
442
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
443
+ checked_at: "2026-10-03"
444
+ unit.pointer_count:
445
+ fill: auto
446
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
447
+ checked_at: "2026-10-03"
448
+ unit.pointer_level:
449
+ fill: judgment
450
+ basis: "each pointer is one tar.gz or zip of one system's logs (Mac, Linux, Zookeeper, OpenStack, Hadoop, HPC and others)"
451
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
452
+ sample_n: 9
453
+ checked_at: "2026-10-04"
454
+ unit.state: {fill: derived, basis: "packed when pointer_level is archive, else expanded"}
455
+ unit.usable_unit:
456
+ fill: judgment
457
+ basis: "8 of 9 archives hold one .log; OpenStack holds 3 logs and anomaly_labels.txt (#05); Hadoop holds 1035 container logs and abnormal_label.txt (#08)"
458
+ files: ["#00", "#01", "#03", "#04", "#05", "#06", "#07", "#08", "#09"]
459
+ sample_n: 9
460
+ checked_at: "2026-10-04"
cards/mastodon-public-timelines.yaml ADDED
@@ -0,0 +1,296 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: mastodon-public-timelines
8
+ parent: null
9
+ title: "Mastodon public posts"
10
+ description: "Public posts from the local public timelines of infosec.exchange, fosstodon.org, hachyderm.io, sigmoid.social, listed newest first through the Mastodon API. One pointer per post; the collection is the instance host name. The hash covers the post's HTML text (content) as returned by the single-post API, as UTF-8 bytes. Each media attachment of a post gets its own pointer to the attachment's original address. Language, tags, reply links, counts and times are in source_meta."
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ audit: {status: draft, reviewed_by: null}
17
+ provenance:
18
+ url: null
19
+ licence: "Per instance: no content license stated; one instance's terms forbid scraping"
20
+ licence_url: "https://infosec.exchange/terms-of-service"
21
+ licence_statement: "Posts belong to their authors; instance terms grant a license to the instance operator only. infosec.exchange: the terms of service (linked) forbid scrapers and data mining tools, so these posts are marked training_ok no. fosstodon.org, hachyderm.io, sigmoid.social: no terms of service and no content license published, so these posts are marked unknown."
22
+ per_collection: null
23
+ training_ok: unknown
24
+ redistribution_ok: unknown
25
+ attribution: null
26
+ per_file: "license_override gives the terms of the post's instance, with notes quoting or describing them."
27
+ licence_tiers: {no/unknown: 1902157, unknown/unknown: 3012770}
28
+ spdx_top: {"(none)": 4914927}
29
+ spdx_distinct: 1
30
+ licence_sample_check: null
31
+ unit:
32
+ pointer_level: null
33
+ pointer_count: 4914927
34
+ units: {"API response": 3770183, file: 1144744}
35
+ state: null
36
+ expanded_count: null
37
+ bytes_total: 1734454627
38
+ rows_without_size: 1144744
39
+ collections:
40
+ distinct: 4
41
+ top: {fosstodon.org: 1447782, hachyderm.io: 1301874, infosec.exchange: 1902157, sigmoid.social: 263114}
42
+ sub_collections: null
43
+ batches: null
44
+ n_ext: 12
45
+ usable_unit: null
46
+ form:
47
+ by_ext:
48
+ html: 3770183
49
+ png: 450883
50
+ jpg: 323380
51
+ jpeg: 274880
52
+ mp4: 81457
53
+ webp: 12445
54
+ mp3: 1208
55
+ gif: 357
56
+ jfif: 109
57
+ webm: 12
58
+ avif: 9
59
+ qt: 4
60
+ bytes_p50: 373
61
+ bytes_p95: 1183
62
+ bytes_max: 64030
63
+ size_quantiles:
64
+ max: 64030
65
+ min: 0
66
+ p25: 173
67
+ p50: 373
68
+ p75: 598
69
+ p90: 917
70
+ p95: 1183
71
+ p99: 1991
72
+ p99.9: 4705
73
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
74
+ by_ext_size:
75
+ - {ext: html, rows: 3770183, bytes: 1734454627, p50: 373, p95: 1183, max: 64030}
76
+ - {ext: png, rows: 450883, bytes: 0, p50: null, p95: null, max: null}
77
+ - {ext: jpg, rows: 323380, bytes: 0, p50: null, p95: null, max: null}
78
+ - {ext: jpeg, rows: 274880, bytes: 0, p50: null, p95: null, max: null}
79
+ - {ext: mp4, rows: 81457, bytes: 0, p50: null, p95: null, max: null}
80
+ - {ext: webp, rows: 12445, bytes: 0, p50: null, p95: null, max: null}
81
+ - {ext: mp3, rows: 1208, bytes: 0, p50: null, p95: null, max: null}
82
+ - {ext: gif, rows: 357, bytes: 0, p50: null, p95: null, max: null}
83
+ - {ext: jfif, rows: 109, bytes: 0, p50: null, p95: null, max: null}
84
+ - {ext: webm, rows: 12, bytes: 0, p50: null, p95: null, max: null}
85
+ - {ext: avif, rows: 9, bytes: 0, p50: null, p95: null, max: null}
86
+ - {ext: qt, rows: 4, bytes: 0, p50: null, p95: null, max: null}
87
+ over_cap:
88
+ cap: 61944279
89
+ rows_over: 0
90
+ bytes_over: 0
91
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
92
+ largest:
93
+ - pointer_id: "mastodon-public-timelines:infosec.exchange:116689048027970676@20260604T003929.764Z"
94
+ size_bytes: 64030
95
+ - pointer_id: "mastodon-public-timelines:infosec.exchange:112163296938462048@20240326T180128.789Z"
96
+ size_bytes: 52956
97
+ - pointer_id: "mastodon-public-timelines:infosec.exchange:109302400309285919@20221107T115750.355Z"
98
+ size_bytes: 49501
99
+ - pointer_id: "mastodon-public-timelines:infosec.exchange:109315565786323307@20221109T194559.581Z"
100
+ size_bytes: 44007
101
+ - pointer_id: "mastodon-public-timelines:infosec.exchange:111152349606769085@20250626T152002.216Z"
102
+ size_bytes: 42586
103
+ date_years:
104
+ key: created_at
105
+ rows_with_date: 4914927
106
+ counts:
107
+ "2017": 2734
108
+ "2018": 20539
109
+ "2019": 41172
110
+ "2020": 57483
111
+ "2021": 69232
112
+ "2022": 475494
113
+ "2023": 1261209
114
+ "2024": 1073245
115
+ "2025": 993657
116
+ "2026": 920162
117
+ media_type:
118
+ "(none)": 113
119
+ audio/mpeg: 1208
120
+ image/avif: 9
121
+ image/gif: 357
122
+ image/jpeg: 598260
123
+ image/png: 450883
124
+ image/webp: 12445
125
+ text/html: 3770183
126
+ video/mp4: 81457
127
+ video/webm: 12
128
+ facets:
129
+ attachment_type: {audio: 1208, gifv: 32113, image: 1062062, unknown: 1, video: 49360}
130
+ instance: {fosstodon.org: 1447782, hachyderm.io: 1301874, infosec.exchange: 1902157, sigmoid.social: 263114}
131
+ media_types:
132
+ audio: 1208
133
+ gifv: 30637
134
+ "gifv,gifv": 207
135
+ "gifv,gifv,gifv": 42
136
+ "gifv,gifv,gifv,gifv": 51
137
+ "gifv,gifv,image": 8
138
+ "gifv,image": 170
139
+ "gifv,image,image": 37
140
+ "gifv,image,image,image": 15
141
+ image: 773560
142
+ "image,gifv": 286
143
+ "image,gifv,gifv": 9
144
+ "image,gifv,image": 15
145
+ "image,image": 59519
146
+ "image,image,gifv": 50
147
+ "image,image,gifv,image": 7
148
+ "image,image,image": 21429
149
+ "image,image,image,gifv": 34
150
+ "image,image,image,image": 26067
151
+ video: 49360
152
+ artifact: null
153
+ source_origin: null
154
+ parse_cost: null
155
+ parse_quirks: null
156
+ structure:
157
+ self_describing: null
158
+ integrity_builtin: null
159
+ has_reference_output: null
160
+ invariants: null
161
+ tools: null
162
+ ships:
163
+ problem_statement: {present: null}
164
+ reference_solution: {share: null}
165
+ tests: {present: null}
166
+ env_definition: {present: null}
167
+ known_answer: {present: null}
168
+ generator: {present: null}
169
+ upstream_labels: {present: null}
170
+ class: null
171
+ subject:
172
+ domain_span: null
173
+ shape_span: null
174
+ skill_family_span: null
175
+ ont_domain: null
176
+ tech_vocabulary: null
177
+ risk:
178
+ contamination: none
179
+ overlap_check:
180
+ done: true
181
+ pointers_flagged: 0
182
+ pointers_removed: 0
183
+ by_class: {}
184
+ benches: {}
185
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
186
+ pii: null
187
+ pii_evidence: null
188
+ malware: null
189
+ ship_bytes: unknown
190
+ subset:
191
+ rule: "take all rows; licence, size and version are labels, not filters"
192
+ rows_in_subset: 4914927
193
+ last_built: null
194
+ access:
195
+ method: "GET uri. Post rows: the response is JSON. Media rows: download the file."
196
+ verify: "Post rows: take the top-level content field, encode it as UTF-8 and compare size and sha256; the stored file is the post's HTML text. Media rows: compute hashes on download."
197
+ hosts:
198
+ "https://cdn.fosstodon.org": 342596
199
+ "https://cdn.masto.host": 104507
200
+ "https://fosstodon.org": 1105186
201
+ "https://hachyderm.io": 1029327
202
+ "https://infosec.exchange": 1477064
203
+ "https://media.hachyderm.io": 272547
204
+ "https://media.infosec.exchange": 425093
205
+ "https://sigmoid.social": 158607
206
+ uri_example: "https://cdn.fosstodon.org/media_attachments/files/000/240/503/original/458b2d259728d320.png"
207
+ needs_auth: {"false": 4914927}
208
+ credential: null
209
+ hash_source: {computed: 3770183, computed_on_fetch: 1144744}
210
+ range_supported: null
211
+ reachable:
212
+ index_flag: {unknown: 4914927}
213
+ index_check: null
214
+ our_sample: null
215
+ drift: null
216
+ sample: null
217
+ evidence:
218
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
219
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
220
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
221
+ class: {fill: unfilled, basis: "ships not judged"}
222
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
223
+ form.by_ext:
224
+ fill: auto
225
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
226
+ checked_at: "2026-10-03"
227
+ form.date_years:
228
+ fill: auto
229
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
230
+ checked_at: "2026-10-03"
231
+ form.over_cap:
232
+ fill: auto
233
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
234
+ checked_at: "2026-10-03"
235
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
237
+ form.size_quantiles:
238
+ fill: auto
239
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
240
+ checked_at: "2026-10-03"
241
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
242
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
243
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
244
+ provenance.licence_tiers:
245
+ fill: auto
246
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
247
+ checked_at: "2026-10-03"
248
+ provenance.spdx_top:
249
+ fill: auto
250
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
251
+ checked_at: "2026-10-03"
252
+ risk.contamination:
253
+ fill: auto
254
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
255
+ checked_at: "2026-10-03"
256
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
257
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
258
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
260
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
261
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
262
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
263
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
264
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
265
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
266
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
267
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
268
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
269
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
270
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
271
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
272
+ snapshot.rev:
273
+ fill: auto
274
+ basis: "last commit listed in the publish receipt of this source"
275
+ checked_at: "2026-10-03"
276
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
277
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ subset:
287
+ fill: auto
288
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
289
+ checked_at: "2026-10-03"
290
+ unit.pointer_count:
291
+ fill: auto
292
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
293
+ checked_at: "2026-10-03"
294
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
295
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
296
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}
cards/mendeley-data.yaml ADDED
@@ -0,0 +1,313 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Source card. Readers: agents that choose sources and fetch files to build terminal tasks.
2
+ # Field names follow the corpus card format where the meaning is the same.
3
+ # evidence.<field path>.fill: auto (computed from the index), judgment (a person read the samples),
4
+ # derived (fixed rule over other fields), unfilled (no sample evidence yet, left empty on purpose).
5
+ # Sample numbers #NN refer to sample.files. Ratios are counts in the sample, not estimates for the source.
6
+ card_version: 1
7
+ source: mendeley-data
8
+ parent: null
9
+ title: null
10
+ description: null
11
+ catalog: terminal-data-pointers
12
+ snapshot:
13
+ rev: null
14
+ at: "2026-10-03"
15
+ rev_note: "No publish receipt for this source yet; rev is empty until it is published."
16
+ header_note: "Still being collected: no collection header yet, so title, description, licence and fetch fields are empty; counts are the pointer shards present at the snapshot date and will grow."
17
+ audit: {status: draft, reviewed_by: null}
18
+ provenance:
19
+ url: null
20
+ licence: null
21
+ licence_url: null
22
+ licence_statement: null
23
+ per_collection: null
24
+ training_ok: unknown
25
+ redistribution_ok: unknown
26
+ attribution: null
27
+ per_file: null
28
+ licence_tiers:
29
+ conditional/conditional: 4119
30
+ conditional/no: 939
31
+ unknown/unknown: 106
32
+ yes/conditional: 1091
33
+ yes/yes: 50232
34
+ spdx_top:
35
+ "(none)": 1045
36
+ Apache-2.0: 459
37
+ BSD-2-Clause: 8
38
+ BSD-3-Clause: 673
39
+ CC-BY-4.0: 45161
40
+ CC-BY-NC-3.0: 4119
41
+ CC0-1.0: 3792
42
+ GPL-2.0-only: 46
43
+ GPL-3.0-only: 1015
44
+ LGPL-2.1-or-later: 30
45
+ MIT: 139
46
+ spdx_distinct: 11
47
+ licence_sample_check: null
48
+ unit:
49
+ pointer_level: null
50
+ pointer_count: 56487
51
+ units: {file: 56487}
52
+ state: null
53
+ expanded_count: null
54
+ bytes_total: 1406885344508
55
+ rows_without_size: 0
56
+ collections:
57
+ distinct: 6245
58
+ top:
59
+ "10.17632/5snrynxwn4.8": 525
60
+ "10.17632/c38dpmg2k2.1": 3284
61
+ "10.17632/dsbywsz3yf.1": 1690
62
+ "10.17632/jwr4x6d3c6.1": 929
63
+ "10.17632/n5r8spjn22.1": 3632
64
+ "10.17632/n7nccp8s9m.1": 3122
65
+ "10.17632/v5hf63zxk2.1": 685
66
+ "10.17632/xfp84r5hb7.1": 854
67
+ "10.17632/xt5bjdhy5g.1": 543
68
+ "10.17632/yhdk6yvh3d.1": 670
69
+ sub_collections: null
70
+ batches: null
71
+ n_ext: 478
72
+ usable_unit: null
73
+ form:
74
+ by_ext:
75
+ tif: 15579
76
+ txt: 5145
77
+ jpg: 3865
78
+ zip: 2815
79
+ xlsx: 2774
80
+ csv: 2354
81
+ gz: 1730
82
+ png: 1373
83
+ mat: 1257
84
+ stk: 1116
85
+ pdf: 1000
86
+ dat: 881
87
+ "(none)": 876
88
+ npz: 856
89
+ hys: 855
90
+ docx: 769
91
+ xls: 591
92
+ tml: 434
93
+ wav: 421
94
+ m: 407
95
+ "(458 other types)": 11389
96
+ bytes_p50: 372308
97
+ bytes_p95: 65497600
98
+ bytes_max: 9050576328
99
+ size_quantiles:
100
+ max: 9050576328
101
+ min: 0
102
+ p25: 49629
103
+ p50: 372308
104
+ p75: 1551124
105
+ p90: 17404474
106
+ p95: 65497600
107
+ p99: 387720911
108
+ p99.9: 3415335321
109
+ method: "nearest rank: sort sizes ascending, take the k-th with k = max(1, ceil(q * n)); rows without a size are left out"
110
+ by_ext_size:
111
+ - {ext: tif, rows: 15579, bytes: 256145637979, p50: 572973, p95: 14017780, max: 9050576328}
112
+ - {ext: txt, rows: 5145, bytes: 4492445082, p50: 47949, p95: 1208474, max: 542869071}
113
+ - {ext: jpg, rows: 3865, bytes: 4692209295, p50: 155420, p95: 7160944, max: 31742842}
114
+ - {ext: zip, rows: 2815, bytes: 499790338982, p50: 4704380, p95: 1030951686, max: 9049417839}
115
+ - {ext: xlsx, rows: 2774, bytes: 2759590834, p50: 48091, p95: 2044464, max: 397320246}
116
+ - {ext: csv, rows: 2354, bytes: 12811103709, p50: 12203, p95: 12184312, max: 1348313176}
117
+ - {ext: gz, rows: 1730, bytes: 141608903305, p50: 412782, p95: 297340248, max: 8527893934}
118
+ - {ext: png, rows: 1373, bytes: 4403619114, p50: 108771, p95: 13593310, max: 165882256}
119
+ - {ext: mat, rows: 1257, bytes: 27808293537, p50: 89377, p95: 105654207, max: 2938901336}
120
+ - {ext: stk, rows: 1116, bytes: 59066040183, p50: 39887123, p95: 132015551, max: 519082065}
121
+ - {ext: pdf, rows: 1000, bytes: 3831075852, p50: 446935, p95: 10380604, max: 1053306128}
122
+ - {ext: dat, rows: 881, bytes: 13337130581, p50: 203574, p95: 138302061, max: 854140000}
123
+ - {ext: "(none)", rows: 876, bytes: 3827123316, p50: 679787, p95: 26214400, max: 486483898}
124
+ - {ext: npz, rows: 856, bytes: 16634700722, p50: 17315936, p95: 44909828, max: 261983165}
125
+ - {ext: hys, rows: 855, bytes: 52363621, p50: 49677, p95: 91106, max: 2372951}
126
+ - {ext: docx, rows: 769, bytes: 1310956155, p50: 35952, p95: 5265392, max: 114070748}
127
+ - {ext: xls, rows: 591, bytes: 419352087, p50: 52736, p95: 1417728, max: 163654656}
128
+ - {ext: tml, rows: 434, bytes: 4963041, p50: 11113, p95: 25466, max: 30687}
129
+ - {ext: wav, rows: 421, bytes: 93031442, p50: 288046, p95: 288046, max: 498252}
130
+ - {ext: m, rows: 407, bytes: 2209353, p50: 2385, p95: 15743, max: 214243}
131
+ over_cap:
132
+ cap: 61944279
133
+ rows_over: 2875
134
+ bytes_over: 1261071149371
135
+ note: "Default fetch-time cap (the overall p99 of benchmark task files, provisional). It only stops a download; it removes no rows from the index or the default subset."
136
+ largest:
137
+ - pointer_id: "mendeley-data:10.17632/zytjrhmpgh.1:0f420efb-3904-48e4-9526-96b59f338424/161013 rpe #7 ncs survival_added drugNCS_100_2c1.tif"
138
+ size_bytes: 9050576328
139
+ - pointer_id: "mendeley-data:10.17632/yztrdmh4cb.1:aa2301ed-b34a-4782-bdbd-2e04152c06b0/161013 rpe #7 ncs survival_added drugNCS_10_2c1.tif"
140
+ size_bytes: 9050576328
141
+ - pointer_id: "mendeley-data:10.17632/wm8p7d6gn8.1:6add337a-62bd-4867-9d2b-35ed1f7dad09/161013 rpe #7 ncs survival_added drugNCS_0_1c1.tif"
142
+ size_bytes: 9050576328
143
+ - pointer_id: "mendeley-data:10.17632/mcfbg64zx2.1:04587376-b4b1-4282-9aba-038b567c0797/161013 rpe #7 ncs survival_added drugNCS_25_3c1.tif"
144
+ size_bytes: 9050576328
145
+ - pointer_id: "mendeley-data:10.17632/k44t49ct74.1:87efaa31-014b-4ba5-9bf8-525d7e4eb84e/161013 rpe #7 ncs survival_added drugNCS_0_3c1.tif"
146
+ size_bytes: 9050576328
147
+ date_years: null
148
+ media_type:
149
+ application/hyperstudio: 1116
150
+ application/octet-stream: 13706
151
+ application/vnd.mendeley-content-ticket.1+json: 999
152
+ application/vnd.openxmlformats-officedocument.spreadsheetml.sheet: 2751
153
+ application/zip: 2421
154
+ image/jpeg: 3906
155
+ image/png: 1366
156
+ image/tiff: 15823
157
+ text/csv: 2294
158
+ text/plain: 5325
159
+ facets:
160
+ license:
161
+ Apache-2.0: 459
162
+ "BSD 2-clause": 8
163
+ "BSD 3-clause": 673
164
+ "CC BY 4.0": 45161
165
+ "CC BY NC 3.0": 4119
166
+ "CC0 1.0": 3792
167
+ "CERN OHL": 3
168
+ CPC: 939
169
+ CeCILL: 4
170
+ GPLv2: 46
171
+ GPLv3: 1015
172
+ LGPL: 30
173
+ MIT: 139
174
+ SPECIAL: 96
175
+ "TAPR OHL": 3
176
+ license_url: {"http://opensource.org/licenses/MIT": 139, "http://www.tapr.org/ohl.html": 3}
177
+ artifact: null
178
+ source_origin: null
179
+ parse_cost: null
180
+ parse_quirks: null
181
+ structure:
182
+ self_describing: null
183
+ integrity_builtin: null
184
+ has_reference_output: null
185
+ invariants: null
186
+ tools: null
187
+ ships:
188
+ problem_statement: {present: null}
189
+ reference_solution: {share: null}
190
+ tests: {present: null}
191
+ env_definition: {present: null}
192
+ known_answer: {present: null}
193
+ generator: {present: null}
194
+ upstream_labels: {present: null}
195
+ class: null
196
+ subject:
197
+ domain_span: null
198
+ shape_span: null
199
+ skill_family_span: null
200
+ ont_domain: null
201
+ tech_vocabulary: null
202
+ risk:
203
+ contamination: none
204
+ overlap_check:
205
+ done: true
206
+ pointers_flagged: 0
207
+ pointers_removed: 0
208
+ by_class: {}
209
+ benches: {}
210
+ method: "pointer names, paths and contents matched against the task files and data addresses of public agent benchmarks; benchmark copies, task files and data addresses are removed, and so are whole containers that copy a benchmark or are derived from one"
211
+ pii: null
212
+ pii_evidence: null
213
+ malware: null
214
+ ship_bytes: unknown
215
+ subset:
216
+ rule: "take all rows; licence, size and version are labels, not filters"
217
+ rows_in_subset: 56487
218
+ last_built: null
219
+ access:
220
+ method: null
221
+ verify: null
222
+ hosts: {"https://data.mendeley.com": 56487}
223
+ uri_example: "https://data.mendeley.com/public-files/datasets/226zz69bk9/files/b8d770e2-4e19-42aa-b69b-a6d5c83fb325/file_downloaded"
224
+ needs_auth: {"false": 56487}
225
+ credential: null
226
+ hash_source: {source_published: 56487}
227
+ range_supported: null
228
+ reachable:
229
+ index_flag: {"(none)": 56487}
230
+ index_check: null
231
+ our_sample: null
232
+ drift: null
233
+ sample: null
234
+ evidence:
235
+ access.credential: {fill: unfilled, basis: "no judgment file for this source yet"}
236
+ access.drift: {fill: unfilled, basis: "no judgment file for this source yet"}
237
+ access.range_supported: {fill: unfilled, basis: "no judgment file for this source yet"}
238
+ class: {fill: unfilled, basis: "ships not judged"}
239
+ form.artifact: {fill: unfilled, basis: "no judgment file for this source yet"}
240
+ form.by_ext:
241
+ fill: auto
242
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
243
+ checked_at: "2026-10-03"
244
+ form.date_years:
245
+ fill: auto
246
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
247
+ checked_at: "2026-10-03"
248
+ form.over_cap:
249
+ fill: auto
250
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
251
+ checked_at: "2026-10-03"
252
+ form.parse_cost: {fill: unfilled, basis: "no judgment file for this source yet"}
253
+ form.parse_quirks: {fill: unfilled, basis: "no judgment file for this source yet"}
254
+ form.size_quantiles:
255
+ fill: auto
256
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
257
+ checked_at: "2026-10-03"
258
+ form.source_origin: {fill: unfilled, basis: "no judgment file for this source yet"}
259
+ provenance.licence: {fill: auto, basis: "collection header licence block", checked_at: "2026-10-03"}
260
+ provenance.licence_sample_check: {fill: unfilled, basis: "no judgment file for this source yet"}
261
+ provenance.licence_tiers:
262
+ fill: auto
263
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
264
+ checked_at: "2026-10-03"
265
+ provenance.spdx_top:
266
+ fill: auto
267
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
268
+ checked_at: "2026-10-03"
269
+ risk.contamination:
270
+ fill: auto
271
+ basis: "benchmark overlap scan over all pointer rows: tb_overlap when a removed pointer matches a Terminal-Bench task, public_benchmark when any pointer is removed, none otherwise"
272
+ checked_at: "2026-10-03"
273
+ risk.malware: {fill: unfilled, basis: "no judgment file for this source yet"}
274
+ risk.pii: {fill: unfilled, basis: "no judgment file for this source yet"}
275
+ risk.pii_evidence: {fill: unfilled, basis: "no judgment file for this source yet"}
276
+ risk.ship_bytes: {fill: derived, basis: "equal to provenance.redistribution_ok"}
277
+ ships.env_definition.kind: {fill: unfilled, basis: "no judgment file for this source yet"}
278
+ ships.env_definition.present: {fill: unfilled, basis: "no judgment file for this source yet"}
279
+ ships.env_definition.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
280
+ ships.generator.present: {fill: unfilled, basis: "no judgment file for this source yet"}
281
+ ships.known_answer.present: {fill: unfilled, basis: "no judgment file for this source yet"}
282
+ ships.problem_statement.present: {fill: unfilled, basis: "no judgment file for this source yet"}
283
+ ships.reference_solution.share: {fill: unfilled, basis: "no judgment file for this source yet"}
284
+ ships.tests.coverage: {fill: unfilled, basis: "no judgment file for this source yet"}
285
+ ships.tests.present: {fill: unfilled, basis: "no judgment file for this source yet"}
286
+ ships.tests.runs: {fill: unfilled, basis: "no judgment file for this source yet"}
287
+ ships.upstream_labels.fields: {fill: unfilled, basis: "no judgment file for this source yet"}
288
+ ships.upstream_labels.present: {fill: unfilled, basis: "no judgment file for this source yet"}
289
+ snapshot.rev:
290
+ fill: auto
291
+ basis: "last commit listed in the publish receipt of this source"
292
+ checked_at: "2026-10-03"
293
+ structure.has_reference_output: {fill: unfilled, basis: "no judgment file for this source yet"}
294
+ structure.integrity_builtin: {fill: unfilled, basis: "no judgment file for this source yet"}
295
+ structure.invariants: {fill: unfilled, basis: "no judgment file for this source yet"}
296
+ structure.self_describing: {fill: unfilled, basis: "no judgment file for this source yet"}
297
+ structure.tools: {fill: unfilled, basis: "no judgment file for this source yet"}
298
+ subject.domain_span: {fill: unfilled, basis: "no judgment file for this source yet"}
299
+ subject.ont_domain: {fill: unfilled, basis: "no judgment file for this source yet"}
300
+ subject.shape_span: {fill: unfilled, basis: "no judgment file for this source yet"}
301
+ subject.skill_family_span: {fill: unfilled, basis: "no judgment file for this source yet"}
302
+ subject.tech_vocabulary: {fill: unfilled, basis: "no judgment file for this source yet"}
303
+ subset:
304
+ fill: auto
305
+ basis: "default subset rule; last_built is the coverage table of the last subset build"
306
+ checked_at: "2026-10-03"
307
+ unit.pointer_count:
308
+ fill: auto
309
+ basis: "computed from every pointer row of this source (licence tiers after the current per-row licence rules)"
310
+ checked_at: "2026-10-03"
311
+ unit.pointer_level: {fill: unfilled, basis: "no judgment file for this source yet"}
312
+ unit.state: {fill: unfilled, basis: "pointer_level not judged"}
313
+ unit.usable_unit: {fill: unfilled, basis: "no judgment file for this source yet"}