Skill Judge

This evaluates your SKILL.md files against official specs and patterns from 17+ reference examples. It scores across dimensions like knowledge delta (does this add expert-only knowledge or just explain what Claude already knows?), thinking patterns versus generic procedures, and anti-pattern quality. The philosophy here is sharp: skills are knowledge externalization, not tutorials, and your job is maximizing expert content while ruthlessly cutting redundancy. Use it when you're writing or reviewing skills and want structured feedback on whether you're actually adding value or just burning tokens on basics. It's opinionated about the difference between educating an AI and wasting context window.

Install to Claude Code

npx -y skills add davila7/claude-code-templates --skill skill-judge --agent claude-code