agent-efficiency
Open-source AI agent efficiency platform: LLM-free context filter + early-exit supervisor cuts token costs ~30% with zero accuracy loss. Works with LangChain, AutoGen, CrewAI, LangGraph and Smolagents.