Skill Evaluator

概要

Ensures agent skills are built for maximum effectiveness by auditing them against Anthropic’s official design patterns and best practices. It combines automated validation scripts with a comprehensive manual scoring rubric that assesses naming conventions, description quality, and structural integrity. By identifying common anti-patterns and providing actionable improvement reports, it helps developers refine their skills for better performance and consistency across various AI agent platforms.

主な機能

• Standardized reporting with scores and prioritized recommendations
• Weighted scoring rubric for naming, description quality, and content
• 68 GitHub stars
• Automated validation of SKILL.md structure and YAML frontmatter
• Guidance on calibrating degrees of freedom for specific task types
• Detection of common anti-patterns like nested references and backslashes

ユースケース

• Identifying content bloat and vague activation triggers in agent instructions
• Auditing a new skill before publishing it to a directory or repository
• Reviewing existing skills for compliance with official Anthropic guidelines