Mega Eval Brief

Generate eval.yaml configuration for the agent eval harness by analyzing a skill's SKILL.md, sub-skills, scripts, and test cases. Useful for setting up evaluation, testing, quality checks, and benchmarking skills.

Tests and benchmarks Claude Code skills empirically via evaluation-driven development. Compares skill vs baseline performance using pass rates, timing, token metrics in quick workflow or 7-phase full pipeline.

Creates new Claude Code skills from scratch, modifies and improves existing ones, runs evaluations and benchmarks performance with variance analysis, optimizes trigger descriptions.