{"skill_id":"ecc-agent-benchmarking","manifest":{"name":"Agent Benchmarking","description":"Head-to-head coding agent comparison tool: YAML task definitions with judge criteria, git worktree isolation per agent run, pass rate/cost/time/consistency metrics, and reproducible benchmarking across Claude Code, Aider, Codex, and other agents.","source_url":"https://github.com/affaan-m/everything-claude-code/tree/main/skills/agent-eval","source_pin":null,"manifest_hash":"9bb29a8b037ffd9dd19de60289d7292858c4c5cc8688632eaa09e9be2e9c7d25","risk_tier":"T2"},"sbom":null,"receipt":null,"root_pubkey":null}