메뉴
Audits every loaded standard module for criterion counts, URL validity, level distribution, and equivalentTo reciprocal consistency. Catches silent drift between a standard's published version and our in-tree module.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | standards-audit |
| description | Audits every loaded standard module for criterion counts, URL validity, level distribution, and equivalentTo reciprocal consistency. Catches silent drift between a standard's published version and our in-tree module. |
| allowed-tools | Bash(bun *) Bash(git *) |
Deep audit of src/standards/ correctness.
criterion.url resolves. Run scripts/check-docs-links.ts over the union.A.equivalentTo: [B], verify B exists in its standard. A reference to a removed criterion is a silent bug.criteria array (guards against typos in the barrel).bun scripts/validate-standards.ts — runs all six checks and emits structured output.standard-builder or type-smith.standards_audited: [wcag22, wcag21, section508, en301549]
checks:
criterion_counts: PASS
level_distribution: PASS
url_validity: PASS (302 URLs checked)
reciprocal_equivalentTo: PASS
duplicate_local_ids: PASS
orphan_criteria: PASS
overall: PASS
Parallel-track dispatcher. Reads .claude/backlog.md, picks the next unchecked item from each of up to 3 active tracks, dispatches them in parallel to specialist subagents, verifies, commits, and checks off. The primary entry point for autonomous ra11y development.
Scaffolds a new accessibility standard module (metadata, criteria, standard export, tests, kb entry). Delegates the criteria enumeration to spec-researcher + standard-builder. Use when adding any conformance framework.
Scaffolds a new ra11y rule end-to-end from a criterion ID using the evaluator-optimizer pattern — spec-researcher produces WCAG context, rule-implementer generates, a11y-reviewer critiques, iterate up to 3 times, then code-reviewer signs off. Use when adding any new rule.
Universal preflight — runs the full check sequence (typecheck, lint, tests, zero-dep, cycles, network isolation, limits, tsdoc, mermaid, kb drift) through the single `bun run verify` entrypoint. Use before every commit and before handing off to another agent.
Scaffolds a new output formatter — file in src/output/formatters/, snapshot tests, CLI --format registration, and a short doc stub. Use when adding terminal/json/sarif/junit/html/markdown/plain or a custom format.
Runs scripts/bench.ts against the performance budget and reports the result. Fails if any scenario exceeds budget. Appends a row to docs/performance.md on success.