skill-deep-auditlisted

# skill-deep-audit — Generic Skill Auditor A read-only, multi-dimensional quality auditor for agent skills. Runs static analysis + optional dryRun reachability checks and produces a scorecard. - **Version**: 1.0.3 - **License**: MIT - **Author**: Evan Song · [github.com/Songhonglei](https://github.com/Songhonglei) - **Repository**: https://github.com/Songhonglei/build-better-skills - **Part of**: [`build-better-skills`](https://github.com/Songhonglei/build-better-skills) suite — see [Stages](https://github.com/Songhonglei/build-better-skills#stages) for the lifecycle map. ## Design principles ``` Can it run? → D3 Portability + D4 Usability conventions Does it run correctly? → D1 Process closure + D6 Code & doc quality Is it safe to run? → D5 Security & op risk Is it well-conformed? → D2 Tool & command conventions Is the whole healthy? → D7 Dependency & footprint ``` ## NEVER DO - Do **not** execute any write operation of the audited skill (read-only, reachability checks, and static analysis only). - Do **not** modify the audited skill's files (read-only and report-only; `--fix` is the one exception and requires explicit user authorization). - Do **not** fabricate check results — when undecidable, mark "cannot confirm, manual verification needed". ## MUST DO - At the start of every audit, **ask the user to choose the check depth (L1 / L2 dryRun) and wait** for the choice. - Any ERR → final result is FAIL, regardless of total score. - After the