self-audit
Skill73 starsChanged 4 months agoMITView on GitHub
Run the verify-plan checklist against YOUR OWN plan before finalizing it (before ExitPlanMode), with special emphasis on the consumer audit. Triggers on: 'audit yourself', 'self-audit', 'check your own plan', 'did you check everything', 'verify before exit plan', or automatically before any non-trivial ExitPlanMode. Catches the gap where a rule exists but was not applied at the decision point.
6 other projects use a version of this.
Adopted by
Most recently taken up by
- helmut-hoffer-von-ankershoffen/heinzeltheir own version, first seen Nov 8, 2026
- 0xUrsanomics/utopia-ostheir own version, first seen Nov 1, 2026
- daichi-0818/claude-code-ops-kittheir own version, first seen Oct 15, 2026
- poa-box/poa-clitheir own version, first seen Jul 31, 2026
- Xxt-XN/agent-self-audittheir own version, first seen Jun 15, 2026
- jamestexas/agentstheir own version, first seen May 16, 2026
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- 0xUrsanomics/utopia-os★ 2same name
- helmut-hoffer-von-ankershoffen/heinzel★ 2same name
- jamestexas/agents★ 2same name
- poa-box/poa-cli★ 2same name
- Xxt-XN/agent-self-audit★ 1same name
- daichi-0818/claude-code-ops-kit★ 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 repository21 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- memory/CLAUDE.mdCLAUDE.md
- changelogSkill
- db-checkSkill
- deploy-checkSkill
- e2e-checkSkill
- elicit-requirementsSkill
- load-sessionSkill
- pr-reviewSkill
- register-projectSkill
- review-diffSkill
- review-fullSkill
- save-diarySkill
- save-sessionSkill
- scaffold-projectSkill
- session-endSkill
- sprint-planSkill
- statusSkill
- tech-radarSkill
- test-checkSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 12
Automatically: before every non-trivial ExitPlanMode (per the Plan-First rule in CLAUDE.md).
- Distinctive: line 13
On demand: when the user says "audit yourself", "did you check everything", and similar.
- Distinctive: line 30
Every hit is a consumer. Ask: "does the plan repoint or update this, or will it break silently?" If a consumer is not handled in the plan, …
“Common” is measured: kept by at least 6 in 10 of the 6 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
Self-audit YOUR OWN plan before ExitPlanMode. This is an attention-gap backstop: the verify-plan checklist already exists in this blueprint (the verify-plan agent), but knowing a check and applying it at the right moment are different things. This skill forces the full checklist to run against the plan you just wrote, catching blind spots you missed while heads-down in planning.
When this runs
- Automatically: before every non-trivial
ExitPlanMode(per the Plan-First rule in CLAUDE.md). A plan is non-trivial if it touches more than one file, changes behavior, or modifies config or a shared resource. - On demand: when the user says "audit yourself", "did you check everything", and similar.
Skip only for genuinely trivial plans (a single-line typo fix).
What it does
- Read the current plan: the plan file named in the plan-mode system message.
- Run the verify-plan checklist against the plan, IN ORDER. Do not re-derive it: spawn the
verify-planagent (or readagents/verify-plan.mdand follow its checklist) so the same checks that would review someone else's plan now review yours. Report a table: Check | Status (PASS / FAIL / UNCLEAR) | Finding. - Emphasis on the Consumer Audit, the highest-value and most-missed check:
- Identify every SHARED RESOURCE the plan changes: a token or credential, an env key, a file path, a DB table, an API field, or any string referenced from more than one place.
- For each, grep BOTH the target repo AND your own global config, because a repo-only search misses consumers that live in your personal setup:
# Repo: grep -rn "<resource>" . # Your own global config (adjust to your layout): grep -rn "<resource>" ~/.claude/skills ~/.claude/agents ~/.claude/hooks ~/.claude/rules - Every hit is a consumer. Ask: "does the plan repoint or update this, or will it break silently?" If a consumer is not handled in the plan, that is a FAIL: add it to the plan before exiting.
- This is the exact class of miss where a plan updates a shared value in the repo but leaves stale references in the user's own global config (a repo-only audit reports "zero consumers" while several live outside the repo).
- On any FAIL: fix the plan file first, then re-run the failed checks. Only proceed to ExitPlanMode when the audit is clean, or the remaining items are explicitly accepted or deferred with a reason.
Output
A Check | Status | Finding table, then one line: AUDIT CLEAN: safe to ExitPlanMode, or N FINDINGS: fixing plan before exit (with the fixes applied to the plan file).
What this does NOT do
- It does not review a code diff (a diff review is a separate concern, after the plan is executed).
- It does not run tests (that is post-execution verification).
- It does not duplicate the checklist: it runs the
verify-planchecks as-is.
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.

