agent toolingby cowork-os
Batch
cowork-os/batch
Plan and execute parallel, repeatable migrations across code, docs, research artifacts, ops configs, or mixed task sets.
What it changes
Passed 14 of 22 test cases with the skill, and 11 of 22 without it.
Without skill50%
With skill64%
Checks it passed that the model alone missed
- Cap maximum parallel workers at 8
- Keep legacy queue and review skills active
- Cap worker concurrency at 8 parallel workers in the plan
and 1 more in the full report
4 fixed10 passed either way7 still failing1 broke
- Pass-rate lift
- +14 pp
- Tokens
- -20%
- Agent turns
- 0%
- Response time
- -42%
Model gemini-3.6-flashMeasured 8/20/2026v1 · latest is v214 of 22 cases passed with skill
This benchmark measured an earlier version of the skill.
See it in action
One of 4 test cases the skill fixed. Same model, same prompt, with and without the skill.
PromptWe are introducing a broad multi-repository refactoring job. Should existing background queue skills and review workflows be deactivated?
What was checked, without and with the skill| Checked | Without skill | With skill |
|---|
| Keep legacy queue and review skills active | Missed | Passed |
|---|
What's inside
The instructions your model reads when the skill loads.
SKILL.md46 lines2 files
Batch
Purpose
Plan and execute repeatable work across many independent targets in parallel.
Routing
- Use when: Use when the user asks to run parallelizable migrations or repeated transformations across many independent targets in any domain.
- Do not use when: Do not use for single-target quality polishing (use simplify) or when the request is purely conceptual with no execution intent.
- Outputs: A parallel execution plan, per-target outcomes, validations, and explicit next actions.
- Success criteria: Work is partitioned, executed with bounded parallelism, validated per target, and summarized with clear blockers/safety boundaries.
Trigger Examples
Positive
- Use /batch migrate src/ from Solid to React.
- Run /batch to update all policy docs to the new template.
- Batch-convert these reports and validate each output.
- Apply this config migration across all environments in parallel.
Negative
- Clean up this one function for readability.
- Improve this single memo's tone and clarity.
- Discuss migration options without executing any changes.
- Use simplify for one-off quality tuning instead of batch.
Parameters
Guidance
Track record
Benchmark
14 / 22
test cases passed, vs 11 / 22 without
Adoption
- Users
- 0
- Times used
- 0
Safety and source details
SkillSafety
✓ 3 of 3 passed✓Code scan15 of 15 checks clean
0 critical · 0 warning · 0 info — scanner v5 — 9/25/2026
View 15 scanner checks
live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing
✓IntentSafe
LLM judge · does the skill do what it claims
✓ContentClean
LLM judge · policy & harmful-content review
Source
ImportedLicense MITImported 9/25/2026Version v2Bundled 1 file