メニュー
Systematic code quality audit. Report only — never fix.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Multi-agent orchestration — delegates to domain experts for complex tasks
Triage and act on the weekly drift-guard rolling issue (transitive dependency drift)
Version bump and release checklist for remote-store
Triage and act on the weekly [mutation] rolling issue (mutation-testing run health)
Read review comments from a PR, fix each issue, resolve threads, validate
Create a pull request for the current branch
| name | audit |
| description | Systematic code quality audit. Report only — never fix. |
| disable-model-invocation | true |
| argument-hint | [scope] |
$ARGUMENTS — what to audit (e.g., "extensions package", "error handling in S3 backend", "test coverage for Azure", "docs accuracy for SFTP guide"). Ask if missing.
Confirm with the user: target (which files — code, tests, docs, specs, any project files), focus (bugs/security/spec/performance/consistency/docs-accuracy/all), depth (quick scan or line-by-line). If $ARGUMENTS answers all three, state your interpretation and ask for confirmation.
Read the target files. For each finding record: file:line, category (Bug/Security/Spec/Consistency/Performance/Docs), severity (High/Medium/Low), evidence.
Docs findings: apply sdd/CONTENT-RULES.md; cite the rule number in the evidence.
## Audit: <target> — N findings
| # | Location | Category | Severity | Summary |
|---|----------|----------|----------|---------|