AGENT SKILL

skill-redaction-audit

暂无简介

License 许可
MIT
最近更新
2026-10-24
Stars
★ 0
Forks
0

02 资源文件 24

03 SKILL.md 正文

Skill Redaction Audit

Use this skill before publishing or sharing an agent-skill bundle, generated skill fixture, release candidate, or public runbook.

Required Inputs

  • A local path to the skill repo or bundle.
  • Optional .redaction-allowlist.json for intentional fake examples.
  • Optional generated path prefixes to exclude from the source-bundle audit.

Side-Effect Boundaries

This skill is read-only. It scans local files and produces a report. It must not rewrite files, upload artifacts, call external accounts, or approve publication.

Approval Requirements

Ask the user before ignoring an error finding. Ask for explicit approval before publishing a repo that still has warning findings involving real-looking people, customer names, or live-action language.

Inline redaction-audit-ignore-next-line comments are allowed only for deterministic fake fixtures. Treat any suppressed finding count as a review item, and do not use suppressions for values copied from private workspaces or live accounts.

Examples

skill-redaction-audit scan ./my-skill --format markdown
skill-redaction-audit scan ./my-skill --allowlist ./.redaction-allowlist.json --fail-on warning
skill-redaction-audit scan ./my-skill --exclude generated --exclude coverage

Validation

Run npm run validate. Attach the audit summary to release-candidate PRs for public skill repos.

04 同仓库更多 Skills