Borrowing it
Nothing to install: this file belongs to jakubsuplicki/codument. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/jakubsuplicki/codument/main/.agents/skills/review-work/SKILL.mdgit clone --depth 1 https://github.com/jakubsuplicki/codumentWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/jakubsuplicki/codument/review-work)<a href="https://agentmods.dev/skills/jakubsuplicki/codument/review-work"><img src="https://agentmods.dev/badge/skills/jakubsuplicki/codument/review-work/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/jakubsuplicki/codument/review-work"><img src="https://agentmods.dev/badge/skills/jakubsuplicki/codument/review-work.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00024 | $0.01953 |
| Opus 5.5 | $0.00010 | $0.00781 |
| Sonnet 5.5 | $0.00005 | $0.00391 |
| Haiku 4.5 | $0.00002 | $0.00195 |
Grade A, and why
review-work scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 23d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 127 lines — stays where its author put it; the contents beside it link to each section on GitHub.
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.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 23d ago Changed · +29 lines 11286b137796
- 1mo ago Changed · +10 lines f4fed02e0da9
- 1mo ago Changed · -3 lines aaa175932b32
- 1mo ago First seen · 91 lines · 24 tokens per session scan A ad7ad0eb0c49
review-work is a skill published in the GitHub repository jakubsuplicki/codument (48 stars, last pushed 23d ago), licensed Apache-2.0. It adds 24 tokens to every session and 1,953 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
review
Adversarial review of an increment before it ships, by a fresh context or another model family; every finding cites path:line and is re-verified. Use for "review", "qa", "critique this".
sw-review
Adversarial review of an increment before it ships, by a fresh context or another model family; every finding cites path:line and is re-verified. Use for "review", "qa", "critique this".
kn-review
Use when reviewing implemented code before committing — multi-perspective review with severity-based findings.
kn-doc
Use when working with Knowns documentation - viewing, searching, creating, or updating docs.
receiving-feedback
Use when receiving code review feedback or corrections - before implementing suggestions. Requires technical verification and reasoned evaluation, not blind agreement or performative responses.
atlas-review
Review code changes and capture learnings for CLAUDE.md by following the canonical review command doc.