#tool-lockout

Cross-lab Routing 기반 3-Agent Blind Eval로 LLM self-preference 한계 극복

I open-sourced a 3-agent blind eval team. Any agent runtime can call it for pre-commitment review of its own plans.