Evaluates and audits AI Agent Skill design quality against official specifications and expert best practices.
Skill Judge is a specialized diagnostic tool designed to review, score, and improve SKILL.md files for AI coding agents. It utilizes a 120-point multi-dimensional framework to assess the 'Knowledge Delta'—ensuring skills provide high-value expert insights rather than redundant tutorials. By identifying anti-patterns and auditing activation descriptions, it helps developers eliminate token waste and refine agent thinking patterns for maximum performance.
Key Features
01Knowledge Delta analysis to minimize redundant token usage
02Comprehensive frontmatter and trigger description auditing
03Anti-pattern identification and 'NEVER' list optimization
04Expert mindset and domain-specific procedure validation
05Multi-dimensional 120-point scoring system for skill quality
06347 GitHub stars
Use Cases
01Auditing new agent skills before deployment to a production toolkit
02Refining existing skill packages to reduce context window consumption
03Standardizing documentation for team-based AI agent development
