Syndicate604/skills-desk-v1

Skills Desk

Frozen v1 copy. Identical to Syndicate604/skills-desk at revision a785ccdc9784ed863d0ffecec826d72176c0a785 (40 training tasks, image ghcr.io/syndicate604/skills-desk@sha256:5492135d…), except for this note. It exists so v1 resubmissions link the dataset they actually train on; Syndicate604/skills-desk main now holds v2.

An OpenEnv environment for the OpenEnv Arena that trains reliable problem solving with tools: understand the request, find the relevant information, compute or repair something, produce the required output, and check it. Each episode is a small workspace under /app with a short instruction. The agent runs commands (exec), writes files (write_file) and sends evaluate; a hidden grader checks the result and returns 1 or 0.

Image: ghcr.io/syndicate604/skills-desk (linux/amd64). Source: source/. Forked from Science Desk and dnakhla/terminal-desk v2 (Apache-2.0); see NOTICE.

Interface

Terminal Desk v2's copy of OpenEnv's upstream tbench2_env contract: exec, write_file, evaluate, close; the instruction repeated in every observation; output not cut; per-task budgets of 4096 completion tokens (model tokens plus every observation after the first), 8192 context tokens and 32 tool calls. The image has Python 3.12 with numpy, scipy, pandas and astropy, plus jq, bc and sqlite3.

Training tasks (40)

manifest/training_tasks.txt, 8 per skill. Each id is one fixed problem, so a GRPO group compares answers to the same question; <id>-i<k> names further fixed instances.

Skill	Tasks
Find and combine evidence	backorders (2 variants), calibration windows, outstanding deliveries, badge access with inherited roles, duplicate customers, bank reconciliation, first release that fixed an issue
Compute from data	channel statistics, rainfall (2 variants), wrapping meter registers, final grades, FIFO stock cost, rolling-median exceedances, loan schedule
Transform structured data	contacts CSV to typed JSON, log lines to JSON, XML to CSV, long to wide, Markdown table to JSON, JSONL sessions, fixed-width report, regional dates to ISO
Repair and verify	toolkit module, broken report script, pagination, log-line parser, unit conversion, business days, deduplication script, sorted log merge
Follow constraints	roster (2 sizes), correct a one-error draft roster, complete a half-agreed roster, cargo choice, course selection, fewest disks, delivery days

Tasks were not chosen by measured difficulty. Some are deliberately straightforward: they train delivering a valid result reliably, alongside tasks that need real search, computation or repair.

Reserved evaluation tasks (20)

manifest/reserved_tasks.txt: four per skill, two fresh applications of a familiar skill and two combining skills, with different file layouts and solution structures. They are served by the same image so a trained checkpoint can be evaluated on them, but they are never trained on and never used to choose or tune the training tasks. The set is frozen (v1); a change would be published as a new version.

tasks.jsonl lists all 60 ids with their split, skill, output path and fixed instruction.

Reward

Binary on evaluate. Content and structure are graded together: correct JSON with wrong values fails, correct values in a malformed file fail, and a CSV row with extra or missing fields fails. Harmless differences pass (key order, CSV quoting, line endings, and either rounding of an exact half where the task states no rounding rule; a task that states half-up rounding requires it). Tasks with many correct answers (rosters, schedules, plans) are graded against their rules. Repair tasks run the repaired code as nobody on hidden input. Nothing rewards a particular tool sequence, reply length or claims of checking.

Verification

Every task: the reference output scores 1, no output and corrupted output score 0, an independent agent-style reference script scores 1 on four instances inside the image, and a deliberate common mistake scores 0. Instances are byte-identical in any process and frozen by fingerprints in source/tests/fixtures/. See source/README.md to build and test.

Limitations

A small bf16 screen of Qwen3.8-27B is a sanity check, not a difficulty estimate; earlier screens underestimated the arena-trained model. Whether training on these tasks transfers can only be measured by evaluating a trained checkpoint on the reserved tasks; the arena's private score is a separate measure.