commit-work
Skill48 starsChanged 40 days agoApache-2.0View on GitHub
Commit an already verified Codument work step with a focused conventional commit.
9 other projects use a version of this.
Adopted by
Most recently taken up by
- imMamdouhaboammar/kaku-chatgpt-harnesstheir own version, first seen Oct 25, 2026
- nesnilnehc/ai-cortextheir own version, first seen Oct 25, 2026
- JayMeDotDot/jmdd-agent-skillstheir own version, first seen Oct 17, 2026
- Samueladewole/claude-skillstheir own version, first seen Oct 13, 2026
- avantmedialtd/skillstheir own version, first seen Oct 9, 2026
- FilenCloudDienste/filen-rstheir own version, first seen Oct 9, 2026
Other versions9 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 9
- jMerta/codex-skills★ 130same name
- bartstc/vite-ts-react-template★ 122same name
- FilenCloudDienste/filen-rs★ 104same name
- nesnilnehc/ai-cortex★ 8same name
- avantmedialtd/skills★ 1same name
- JayMeDotDot/jmdd-agent-skills★ 1same name
- Rafaelrr5/agent-cortex★ 1same name
- imMamdouhaboammar/kaku-chatgpt-harness★ 0same name
- Samueladewole/claude-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository31 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- code-reviewerSkill
- establish-charterSkill
- frontend-designSkill
- grill-with-docsSkill
- motion-craftSkill
- plan-with-docsSkill
- review-codebaseSkill
- review-workSkill
- senior-architectSkill
- senior-backendSkill
- senior-frontendSkill
- tddSkill
- update-docsSkill
- work-stepSkill
- code-reviewerSkill
- commit-workSkill
- establish-charterSkill
- frontend-designSkill
- grill-with-docsSkill
- motion-craftSkill
- plan-with-docsSkill
- review-codebaseSkill
- review-workSkill
- senior-architectSkill
- senior-backendSkill
- senior-frontendSkill
- tddSkill
- update-docsSkill
- work-stepSkill
The file
We split this skill into the rules it gives an agent
12 rules. 0 are common, 8 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 9 edited versions we read keep these.
8
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Use this after review-work is clean, or after safe contracted fixes and user-approved fixes are resolved and the user has explicitly …
- Very specific: line 14
Check that review-work is clean, or that every finding was fixed or explicitly deferred by the user.
- Distinctive: line 15
Check git status --short and git diff --cached.
- Very specific: line 17
Commit that boundary with a conventional commit prefix.
- Distinctive: line 40
Do not commit unresolved high or critical review findings.
- Distinctive: line 41
Do not commit without a passing receipt for the exact staged boundary.
- Distinctive: line 42
Do not decide to defer review findings yourself.
- Distinctive: line 43
Do not stage or commit unrelated dirty files.
- Distinctive: line 44
Do not claim verification passed if a command failed or could not run.
- Very specific: line 45
Commit as the user only. Never add a Co-Authored-By trailer for the AI agent (for example Claude or Codex), in any profile.
- Distinctive: line 46
Follow any repository-specific commit timestamp or signing rules from AGENTS.md, CLAUDE.md, or the active feature plan.
- Distinctive: line 48
Do not start the next delivery-plan step as part of compacting context.
“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Commit Work
Use this after review-work is clean, or after safe contracted fixes and user-approved fixes are
resolved and the user has explicitly deferred any remaining finding.
Workflow
- Check the active plan step is complete, using its recovery copy at
.codument/pending-plans/<repo-relative-plan-path>after final-step compaction. For the final step, confirm the original doc's## Delivery Planblock has been compacted out (plan-with-docs → Compaction on ship) — a shipped feature doc must not commit with a stale delivery checklist still in it. - Check that
review-workis clean, or that every finding was fixed or explicitly deferred by the user. - Check
git status --shortandgit diff --cached. - Confirm
codument work status --jsonshows the reviewed step ready and the already-staged, verified boundary unchanged; do not stage here. If it changed, return to review and verification. Respect an explicit no-commit request: leave the step ready and stop. Commit that boundary with a conventional commit prefix. The managed pre-commit hook runs
codument verify; when the staged boundary, approval, review policy and review evidence remain current, it reuses the exact receipt instead of repeating the review:feat:fix:docs:test:refactor:chore:
A routine hook failure returns to diagnosis, correction and exact staged review within the approved contract. Never bypass the hook or commit a red boundary; pause only if repair needs an unapproved decision, external input or genuinely irreversible action, or the user has requested a pause.
- If that was the plan's last step, remove only this plan's recovery copy after confirming the commit succeeded, then run
codument doctor --strictonce and report what it says. If interrupted before cleanup, reconcile the copy with the committed boundary before resuming; a leftover copy does not authorize repeating completed work. Doctor reports; it does not gate — a plan must not be blocked by an adopting repo's pre-existing debt. This is the only moment the loop looks at repo-wide health: stagedverifyanswers whether this change is ready,doctoranswers whether the knowledge base is still worth reading. Once per plan, not once per step. After the successful commit, run
codument work finishto reconcile delivery, then readcodument steps --plan <active-plan> --jsonif the Delivery Plan remains. Name the next step and continue towork-stepunless gated or single-step mode requires stopping. Completed local state must agree with committed delivery; a missing checklist alone is not completion. In gated mode, fill the next-step gate from that readback:Step N is reviewed and committed. Next options: 1. Start Step [next ordinal]: [actual next step text] with /work-step 2. Review the plan before continuing 3. Compact context before continuing 4. Pause here- If the user chooses compact context, write the Resume checkpoint inside the active Delivery Plan before using a native context-compaction command. If none exists, provide that concise restart note grounded in the plan, mapped docs, and
git status, then pause. On a completed plan, give a completion note rather than recreating delivery scaffolding.
Rules
- Do not commit unresolved high or critical review findings.
- Do not commit without a passing receipt for the exact staged boundary.
- Do not decide to defer review findings yourself; only the user can defer findings.
- Do not stage or commit unrelated dirty files.
- Do not claim verification passed if a command failed or could not run.
- Commit as the user only. Never add a
Co-Authored-Bytrailer for the AI agent (for example Claude or Codex), in any profile. - Follow any repository-specific commit timestamp or signing rules from
AGENTS.md,CLAUDE.md, or the active feature plan. - In gated mode, do not start the next delivery-plan step in the same response as the commit.
- Do not start the next delivery-plan step as part of compacting context.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

