eval-bench 是做什么的？
Benchmark Claude Code plugins/skills/agents/MCPs by A/B comparing versions with LLM-judged evaluation prompts
Benchmark Claude Code plugins/skills/agents/MCPs by A/B comparing versions with LLM-judged evaluation prompts
eval-bench is a Claude Code agent skill that benchmark Claude Code plugins/skills/agents/MCPs by A/B comparing versions with LLM-judged evaluation prompts.
npx skills add derberg/eval-benchBenchmark Claude Code plugins/skills/agents/MCPs by A/B comparing versions with LLM-judged evaluation prompts