skills-auditor
Detect incorrectly installed skills and report issues.
When & Why to Use This Skill
The skills_auditor is a specialized Claude skill designed to maintain the integrity of an agent's ecosystem by detecting incorrectly installed, broken, or missing Codex-native skills. It automates the validation of skill metadata and directory consistency, providing detailed reporting and safe fix proposals to ensure a stable and reliable AI agent environment. By monitoring the health of the agent's toolset, it prevents runtime errors and improves overall system maintainability.
Use Cases
- Skill Integrity Auditing: Automatically validate .codex/skills frontmatter and folder structures to identify configuration errors before they cause agent failures.
- Environment Synchronization: Compare skill directories with documentation folders to ensure that all agent capabilities are properly synced and documented.
- Automated Health Reporting: Generate comprehensive status logs in SKILLS_STATUS.md to provide developers with immediate visibility into the health and availability of the agent's skills.
- Safe Configuration Recovery: Propose corrective actions for broken skills via a merge protocol, allowing for controlled fixes without the risk of blind overwrites.
| name | skills_auditor |
|---|---|
| description | Detect incorrectly installed skills and report issues. |
| short-description | Skills audit |
Purpose
Find broken or missing Codex-native skills.
Steps
- Validate
.codex/skills/*/SKILL.mdfrontmatter and folder names. - Compare
.codex/skills/with.agent-docs/skills/. - Record findings in
.agent-docs/memory/SKILLS_STATUS.md.
Guardrails
- Propose fixes via merge protocol, do not overwrite blindly.