This is quality control for your Claude skills. It loads a skill's SKILL.md, checks it against documented best practices across five dimensions (structure, description, workflow, token efficiency, anti-patterns), and returns concrete fixes, not vague advice. The iron law is brutal: never say "could be improved" without stating exactly what to change. Use it when you've written a skill and want to know if your workflow actually guides the model or if your description is too abstract to route properly. The review format leads with what works (quoted, specific) before prioritized suggestions with before/after examples. Think of it as a linter that understands prompt engineering principles, not just syntax.
npx skills add https://github.com/sanyuan0704/sanyuan-skills --skill skill-review