Headroom is a local-first context compression layer that reduces token usage for AI agents by compressing tool outputs, logs, files, and RAG chunks before they reach the LLM. It offers library, proxy, and MCP server modes and claims to preserve answer accuracy while cutting tokens by 15-95% depending on workload.
Serena is an MIT-licensed, Python-based MCP toolkit that provides AI coding agents with IDE-grade semantic code retrieval, editing, refactoring, and debugging capabilities through high-level symbol-level abstractions.
