from claude-code-ultimate-guide3,149
Run a quantitative quality audit on agents, skills, and commands to check production readiness and get prioritized recommendations.
Performs an automated quality audit for Claude Code agents, skills, and commands using a 16-criteria scoring framework. Produces JSON and Markdown reports with per-file scores, grades, and prioritized fix suggestions to bring items up to production readiness.
Run this skill before releases, PR reviews, or as part of CI to prevent low-quality agent deployments. Use the quick mode for fast checks and full/comparative modes for deeper analysis.
Designed for Claude Code environments and CI pipelines; outputs JSON suitable for automation and gating in GitHub Actions or other CI systems.
A comprehensive Claude Code skill for auditing agents, skills, and commands against a 16-criteria quality framework. The SKILL.md is very well-documented with detailed methodology, scoring rubrics, output examples, and CI/CD integration guidance. No bundled scripts — it's a purely instructional skill relying on the agent to execute the audit logic. Useful for Claude Code users but niche audience; limited applicability outside that ecosystem.
Well-structured and thorough documentation. Security is good — no curl|bash, no hardcoded creds, no exfiltration risks. Minor concern: the skill instructs agents to read project files broadly, but this is inherent to an audit tool. Code quality is decent but the skill has no executable scripts to verify correctness — it's a long instructional document. Architecture is good with clear phases and output contracts, though it references external files (criteria.yaml) not included. Usefulness is moderate — solves a real problem for Claude Code users specifically, but the audience is narrow and it requires manual implementation of the scoring engine.
