skill-reviewer

To audit AI skill directories, you can review SKILL.md, guidelines.md, and commands files across eight dimensions like orchestration and schema consistency, classifying findings by severity and writing structured reports to .artifacts/skill-reviewer/{skill-name}/review.md.

Workflow validation in AI skills checks for structural integrity, YAML frontmatter consistency, reference integrity, and step sequencing correctness to identify problems before they cause degraded behavior.

Yes, you can check YAML frontmatter and reference integrity by running deterministic pre-review checks across SKILL.md, skills/*.md, and commands/*.md files to produce evidence-based findings.

Skill auditing classifies structural and instructional findings by severity as CRITICAL or HIGH blockers, or MEDIUM and LOW suggestions, requiring each finding to include a concrete improvement.

The best way to validate instruction clarity is through an eight-dimension review evaluating orchestration, step sequencing, schema consistency, cognitive load, documentation alignment, command naming, and error handling.

A skill audit for workflow correctness is not suitable when you only need simple syntax validation, as it comprehensively evaluates eight dimensions including orchestration, routing, and cognitive load across all relevant directory files.