NeVibeCode · 2 items
Skills and MCP tagged “llm-judge”
rewardkit skill A skill for writing Harbor task verifiers with Reward Kit: programmatic criteria, LLM judges and a single reward score built from a tests/ directory.
Claude Code CX Codex Cursor +1
1,300installs
eval skill Ranks every agent's result in an AgentHub session, by a metric command or an LLM judge, then names a winner. Handy once several agents have finished the same task.
Claude Code
1,408installs
