2026-04-25

Added

• headroom - Context optimization layer for LLM apps (1.5k ⭐): compresses tool outputs/logs/RAG/file reads locally and losslessly (CCR), so agents see less noise and use far fewer tokens.