SkillCompass helps developers systematically improve AI agent capabilities by identifying and fixing weaknesses across six performance dimensions.

Evaluates Claude Code and OpenClaw agents across 6 skill dimensions: structure, trigger, security, functional, comparative, and uniqueness

Automatically identifies the weakest skill area and provides fixes, then progressively improves the next lowest-scoring dimension

Detects when model updates make certain skills obsolete, keeping evaluations relevant

Runs locally with Node.js v18+ requirement, giving developers full control over the assessment process