ArcReport commited on
Commit
d3ac8d2
·
verified ·
1 Parent(s): 69836a2

Card: precheck quick-start first

Browse files
Files changed (1) hide show
  1. README.md +44 -13
README.md CHANGED
@@ -1,6 +1,6 @@
1
  ---
2
  license: cc-by-4.0
3
- pretty_name: "ARC: can AI shopping agents buy from these stores?"
4
  language:
5
  - en
6
  tags:
@@ -31,7 +31,28 @@ configs:
31
  data_files: latest/protocol_matrix.parquet
32
  ---
33
 
34
- # 3,538 online stores checked: can AI shopping agents buy from them?
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
35
 
36
  **Snapshot `2026-10-12`: 3,538 stores, checked 6 Oct 2026 to 12 Oct 2026.** The check finished for 3,497 of them (41 did not respond in time and are marked `not_tested`). Percentages below are of those 3,497.
37
 
@@ -125,14 +146,14 @@ No scraped page text or product copy, no model-written prose, no costs, no email
125
 
126
  Mirror of the official ARC open dataset at [arcreport.ai/data](https://www.arcreport.ai/data), updated after each weekly snapshot (Mondays). Files are byte-for-byte copies of ARC's published CSV and Parquet.
127
 
128
- **Current snapshot:** `2026-10-12` (snapshot date 2026-10-12, generated 2026-10-12 13:34 UTC, method `arc-method-2026.10`).
129
 
130
  | Config / table | Rows | What it holds |
131
  |---|---|---|
132
  | `store_answers` | 3,538 | Start here. One row per store in plain words: can AI agents find products and get a checkout link, what blocks them, a one-sentence verdict, top fix, and whether the answer is from the automated HTTP check or also the browser cart test. |
133
  | `stores` | 89 | Latest completed ARC Index result per store (browser cart test: score, personas that carted, failure label). |
134
  | `rails_checks` | 3,538 | Latest HTTP agent check per store domain (UCP profile, WebMCP, storefront MCP, catalog tool, cart link, checkout WebMCP). |
135
- | `adoption_history` | 29 | Weekly aggregate counts by source and metric (long format). |
136
  | `personas` | 5 | The five ARC browser shopper personas (A to E) with legacy alias ids. |
137
  | `protocol_matrix` | 24 | Declared / Valid / Usable / Transacts counts per protocol with evidence labels. |
138
 
@@ -156,17 +177,27 @@ Or just download [`latest/store_answers.csv`](https://huggingface.co/datasets/Ar
156
  - `2026-10-09.r2` (generated 2026-10-09 15:54 UTC, method arc-method-2026.10)
157
  - `2026-10-09.r3` (generated 2026-10-09 16:03 UTC, method arc-method-2026.10)
158
  - `2026-10-12` (generated 2026-10-12 13:34 UTC, method arc-method-2026.10)
 
 
 
159
 
160
- ### Citation
 
 
 
 
161
 
162
  ```bibtex
163
- @misc{arc_agent_commerce_index_2026,
164
- title = {ARC: can AI shopping agents buy from these stores? Open dataset},
165
- author = {{ARC}},
166
- year = {2026},
167
- note = {Snapshot 2026-10-12, method arc-method-2026.10},
168
- howpublished = {\url{https://huggingface.co/datasets/ArcReport/arc-agent-commerce}},
169
- url = {https://www.arcreport.ai/data},
170
- license = {CC-BY-4.0}
 
171
  }
172
  ```
 
 
 
1
  ---
2
  license: cc-by-4.0
3
+ pretty_name: "ARC: can AI agents reach checkout at these stores?"
4
  language:
5
  - en
6
  tags:
 
31
  data_files: latest/protocol_matrix.parquet
32
  ---
33
 
34
+ ## Quick start: precheck (free, no sign-in, CC BY 4.0)
35
+
36
+ Before your agent tries a store, ask ARC whether agents can reach checkout there:
37
+
38
+ ```bash
39
+ curl "https://www.arcreport.ai/api/precheck?url=https://allbirds.com"
40
+ ```
41
+
42
+ Real response (trimmed, 14 Oct 2026):
43
+
44
+ ```json
45
+ {"host":"allbirds.com","known":true,"reachable_checkout":true,
46
+ "rails":[{"name":"UCP","status":"valid"},{"name":"ACP","status":"not_found"},
47
+ {"name":"MCP","status":"yes"},{"name":"Checkout WebMCP","status":"detected"}],
48
+ "blockers":[],"confidence":"high","last_checked":"2026-10-13T09:47:13.000Z",
49
+ "stale":true,"source":"ARC (arcreport.ai)","licence":"CC BY 4.0"}
50
+ ```
51
+
52
+ `reachable_checkout` is based on ARC's last saved check. It does not mean an agent completed checkout. Check `last_checked` and `stale` before relying on it. Contact: arcreport@mail.grokbot.com (ARC).
53
+
54
+
55
+ # 3,538 online stores checked: can AI agents reach checkout?
56
 
57
  **Snapshot `2026-10-12`: 3,538 stores, checked 6 Oct 2026 to 12 Oct 2026.** The check finished for 3,497 of them (41 did not respond in time and are marked `not_tested`). Percentages below are of those 3,497.
58
 
 
146
 
147
  Mirror of the official ARC open dataset at [arcreport.ai/data](https://www.arcreport.ai/data), updated after each weekly snapshot (Mondays). Files are byte-for-byte copies of ARC's published CSV and Parquet.
148
 
149
+ **Current snapshot:** `2026-10-12.r2` (snapshot date 2026-10-12, generated 2026-10-12 20:37 UTC, method `arc-method-2026.10`, supersedes `2026-10-12`).
150
 
151
  | Config / table | Rows | What it holds |
152
  |---|---|---|
153
  | `store_answers` | 3,538 | Start here. One row per store in plain words: can AI agents find products and get a checkout link, what blocks them, a one-sentence verdict, top fix, and whether the answer is from the automated HTTP check or also the browser cart test. |
154
  | `stores` | 89 | Latest completed ARC Index result per store (browser cart test: score, personas that carted, failure label). |
155
  | `rails_checks` | 3,538 | Latest HTTP agent check per store domain (UCP profile, WebMCP, storefront MCP, catalog tool, cart link, checkout WebMCP). |
156
+ | `adoption_history` | 32 | Weekly aggregate counts by source and metric (long format). |
157
  | `personas` | 5 | The five ARC browser shopper personas (A to E) with legacy alias ids. |
158
  | `protocol_matrix` | 24 | Declared / Valid / Usable / Transacts counts per protocol with evidence labels. |
159
 
 
177
  - `2026-10-09.r2` (generated 2026-10-09 15:54 UTC, method arc-method-2026.10)
178
  - `2026-10-09.r3` (generated 2026-10-09 16:03 UTC, method arc-method-2026.10)
179
  - `2026-10-12` (generated 2026-10-12 13:34 UTC, method arc-method-2026.10)
180
+ - `2026-10-12.r2` (generated 2026-10-12 20:37 UTC, method arc-method-2026.10; adds `publishes_ucp_file`, `ucp_file_checked_at` to store_answers and `ucp_file` to rails_checks; adoption_history 29→32 rows)
181
+
182
+ ## Cite as
183
 
184
+ **Cite as: ARC (arcreport.ai).** Please cite a dated version, not `latest/`.
185
+
186
+ > ARC (arcreport.ai). *ARC Agent Commerce Index: can AI shopping agents buy from these stores?* Open dataset, snapshot 2026-10-12.r2, method arc-method-2026.10. CC-BY-4.0. https://www.arcreport.ai/data
187
+
188
+ A machine-readable [`CITATION.cff`](https://www.arcreport.ai/CITATION.cff) is published with the data.
189
 
190
  ```bibtex
191
+ @misc{arc_agent_commerce_2026,
192
+ author = {{ARC (arcreport.ai)}},
193
+ title = {{ARC Agent Commerce Index: can AI shopping agents buy from these stores?}},
194
+ year = {2026},
195
+ version = {2026-10-12.r2},
196
+ note = {Open dataset, method arc-method-2026.10. Licensed CC-BY-4.0},
197
+ publisher = {ARC},
198
+ url = {https://www.arcreport.ai/data},
199
+ howpublished = {\url{https://huggingface.co/datasets/ArcReport/arc-agent-commerce}}
200
  }
201
  ```
202
+
203
+ When a Zenodo DOI exists, add `doi = {10.5281/zenodo.XXXXXXX}` to the entry.