review-work
Skill48 starsChanged 40 days agoApache-2.0View on GitHub
Review the current diff against the approved Codument plan, tests, docs, registry, and architecture boundaries.
8 other projects use a version of this.
Adopted by
Most recently taken up by
- benmix/skillstheir own version, first seen Oct 13, 2026
- wjgoarxiv/litopencodetheir own version, first seen Oct 13, 2026
- AgiFlow/ai-plugintheir own version, first seen Oct 10, 2026
- hassan-mohiddin/freeflowtheir own version, first seen Oct 4, 2026
- djt889/OhMyZcodetheir own version, first seen Sep 1, 2026
- Ostico/Skillstheir own version, first seen Aug 12, 2026
Other versions8 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 8
- agoodway/.claude★ 11same name
- hassan-mohiddin/freeflow★ 5same name
- AgiFlow/ai-plugin★ 4same name
- apocohq/skills★ 4same name
- djt889/OhMyZcode★ 3same name
- wjgoarxiv/litopencode★ 2same name
- benmix/skills★ 1same name
- Ostico/Skills★ 1same 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
- 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
- code-reviewerSkill
- commit-workSkill
- establish-charterSkill
- frontend-designSkill
- grill-with-docsSkill
- motion-craftSkill
- plan-with-docsSkill
- review-codebaseSkill
- senior-architectSkill
- senior-backendSkill
- senior-frontendSkill
- tddSkill
- update-docsSkill
- work-stepSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
15 rules. 0 are common, 8 are distinctive, and 7 are very specific to this project.
0
Common
Most of the 7 edited versions we read keep these.
8
Distinctive
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Use this after a planned step has been implemented and before committing.
- Very specific: line 10
If this is the session's first workflow action, run the compatibility check from work-step before lifecycle commands.
- Very specific: line 22
Read codument work status --json and its approved plan step.
- Very specific: line 24
Run codument verify once. Read the generated reviewPolicy.minimum and reasons.
- Very specific: line 27
When a worksheet is required, complete only its top-level invariantsChecked, findings, and signer fields.
- Distinctive: line 55
When delivering a branch to CI, the staged-step gate and full branch review are separate boundaries.
- Very specific: line 58
After staging the intended final changes (and preparing final approval if the plan was compacted), resolve the intended target branch and …
- Very specific: line 94
If there are no findings, say so clearly and mention any remaining verification risk, then continue to commit-work.
- Very specific: line 108
When a review finding is resolved before commit — fixed (a verifiable change) or explicitly deferred — log it once so codument watch / …
- Distinctive: line 120
Treat work outside the approved outcome and constraints as a finding.
- Distinctive: line 121
Do not focus on formatting that automated tools should handle.
- Distinctive: line 122
Do not manufacture issues.
- Distinctive: line 124
Do not mark review complete for commit until contracted safe fixes or user-approved fixes are made and remaining findings are explicitly …
- Distinctive: line 125
Do not commit until high and critical findings are fixed or explicitly deferred by the user.
- Distinctive: line 126
Do not ask to start the next delivery-plan step after review.
“Common” is measured: kept by at least 6 in 10 of the 7 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
Review Work
Use this after a planned step has been implemented and before committing.
If this is the session's first workflow action, run the compatibility check from work-step
before lifecycle commands. The actual executable must expose work and verify; repair a
mismatch instead of substituting older commands or manual checkpoints.
For compact dead-code cleanup, use the same generated worksheet and independent reviewer. Check the unused-code evidence against exports, dynamic loading, framework entry points and side effects; deletion is never automatically trivial. Do not demand a second plan or a separate review bundle when the worksheet already supplies the reviewed boundary. Historical paths inside the plan remain valid scope; durable Key files and registry pointers must describe what remains.
Review Order
- Read
codument work status --jsonand its approved plan step. After final compaction, use the tracked contract and final-delivery binding indocs/.approvals.json; the pending-plans recovery copy supplies working context only. Inspect the compacted doc in the staged boundary as usual. - Inspect
git diff --cached; that exact staged boundary, not the whole dirty worktree, is the review subject. - Run
codument verifyonce. Read the generatedreviewPolicy.minimumandreasons; it writes.codument/review-worksheet.jsonwhen required focused or adversarial review lacks current coverage. - Diagnose routine test, mapping or documentation failures and correct the cause within the approved contract. Batch missing source registration through
codument map materialize <files...>, inspect any partial failure and retry after correcting it, then restage and rerun the affected checks andcodument verify. The staged boundary checks ownership and doc sync together; do not repeat per-file ownership queries or add mirror prose merely to make it green. Ask the human only when repair needs an unapproved decision, external input or a genuinely irreversible action. - Compare the promised outcome and acceptance criteria with the evidence actually obtained at the relevant user or integration boundary. Name untested assumptions and substitutes; missing required evidence remains a blocker unless the user explicitly changes acceptance. Review the exact staged diff, mapped invariants, tests and generated
reviewContextwhen present with the effort below. - When a worksheet is required, complete only its top-level
invariantsChecked,findings, andsignerfields. Do not alter generated context. Run the exact printedcodument verify --record .codument/review-worksheet.jsoncommand; it records and verifies the same staged boundary in that invocation. - Fix safe, obvious findings that follow the approved contract, restage, and return to step 3. Findings involving public interfaces, privacy, security, migrations, data loss, deletions or dependencies receive stronger review; an obvious contracted fix does not require renewed human approval merely for touching that boundary. Pause for unapproved judgment calls, changed promises or constraints, external input or a genuinely irreversible action. A changed boundary invalidates its earlier review automatically; gated mode still waits for the user's findings decision.
- When
codument verifypasses, runcodument work finishto preserve verified readiness, then continue tocommit-workwithout a routine prompt. If the user prohibited commits, stop with ready work and the pending commit visible. The exact receipt lets the hook confirm unchanged staged bytes. Gated and single-step requests retain their limits. - If review pauses or the user redirects work, preserve the plan path, step, next gate, unresolved finding or failure, and resume condition in the plan's Resume checkpoint (in its recovery copy after final-step compaction). Recheck the staged boundary on return; a changed boundary invalidates earlier review. If fixes restore delivery scaffolding, recompact and restage it before recording the final review.
Review effort
For none, make a one-pass diff self-review without creating a required worksheet.
The generated minimum is a floor, not proof of semantic safety. When it permits focused, use
focused self-review only for an understood, small, local and reversible change. Check the affected
behavior, callers, documented invariants and acceptance evidence; escalate to adversarial review
if the behavior is substantial or understanding is uncertain.
For adversarial, use a fresh independent reviewer without the author's conversation when the
host provides one. Attack correctness, privacy, security, data loss, performance, type safety and
architecture fit. Without an independent reviewer, make the same pass and disclose reduced
independence. Disclose focused self-review as such. Public-interface or documented-contract
changes, privacy, security, risk declarations, migrations, deletions, dependencies and uncertain
analysis retain the stronger floor; neither a small file count nor a "low risk" label can waive it.
Both efforts use the ordinary generated worksheet, named finding confirmation and exact receipt path. Fill only its existing reviewer fields; do not rewrite the generated policy or invent an effort artifact. Every covering reproduced-red finding blocks delivery regardless of effort; changed review evidence must not reuse an outdated pass.
Preparing a branch for CI
When delivering a branch to CI, the staged-step gate and full branch review are separate boundaries.
Branch and portable review retain independent adversarial effort; local focused review cannot
substitute for it.
After staging the intended final changes (and preparing final approval if the plan was compacted),
resolve the intended target branch and run codument review --base <ref> --pending --bundle --full.
Give a fresh independent reviewer that bundle and the full branch diff. Record its findings with the
same --base <ref> --pending selection, then export .codument-review.json with --export and stage
it. Never promote local step receipts into branch coverage. Re-run the normal staged review and
verification before committing; changed bytes reopen whichever boundary they affect. Keep private
findings under .codument/. CI uses --committed --require-review --review-file .codument-review.json
against the same resolved base and rejects missing or stale evidence. A moved target base may require
a new review. Run no remote action unless separately authorized.
Output
Lead with findings ordered by severity:
- Critical
- High
- Medium
- Low
Each finding should include the file and line where possible, the concrete risk, and the smallest safe fix.
Auto-apply only safe, obvious fixes within the approved contract and continue to commit-work
after renewed verification and review. Pause for an unapproved judgment call, external input or
genuinely irreversible action. In gated mode, list findings and stop for the user's decision;
do not fix findings automatically.
In gated mode, end with exactly these next options when findings exist:
Review complete. Findings need a decision:
1. Fix all findings
2. Fix selected findings: [list numbers]
3. Defer selected findings with a reason
4. Pause here / make no changesIf there are no findings, say so clearly and mention any remaining verification risk, then continue
to commit-work. An explicit no-commit request takes precedence: report verified readiness and
the pending commit, then stop without offering a commit action. Otherwise, in gated mode end with
exactly these next options:
Review clean. Next options:
1. Run /commit-work for this reviewed step
2. Run extra verification
3. Pause hereRecord resolved findings (impact ledger)
When a review finding is resolved before commit — fixed (a verifiable change) or explicitly deferred — log it once so codument watch / report can show what the review step caught:
codument emit review --tier <correctness|minor> --resolution <fixed|deferred> [--feature <name>] [--step <n>] [--summary "<one line>"]Tier conservatively: correctness covers safety, security, data-loss, and logic bugs; anything cosmetic is minor. This line is explicitly self-reported, and the watch headline counts only fixed × correctness — so do not inflate it. A finding that was neither fixed nor deferred (dismissed as a non-issue) is not logged.
Rules
- Log each resolved finding once with
codument emit review(fixed or deferred); never log one that was neither. - Review only the staged boundary and restage every review fix before rerunning the verifier.
- Treat work outside the approved outcome and constraints as a finding. Additional implementation files in an outcome plan need routing and ownership repair, not a scope-amendment finding; legacy file-bound plans retain strict scope review.
- Do not focus on formatting that automated tools should handle.
- Do not manufacture issues.
- Auto-apply only safe, obvious fixes, and pause the run for judgment-call findings (see above). In gated mode do not fix findings automatically at all; wait for the user to approve all fixes, select specific findings, defer specific findings, or pause.
- Do not mark review complete for commit until contracted safe fixes or user-approved fixes are made and remaining findings are explicitly deferred by the user.
- Do not commit until high and critical findings are fixed or explicitly deferred by the user.
- Do not ask to start the next delivery-plan step after review.
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.

