◇MCP serverOrchestrationFree
Headroom
Compress tool outputs, logs, files and RAG chunks before they reach the LLM — 60-95% fewer tokens, same answers.
Headroom
Compress tool outputs, logs, files and RAG chunks before they reach the LLM — 60–95% fewer tokens with the same answers. Available as a library, a proxy, and an MCP server, so it drops into the agent context pipeline.
Highlights:
- MCP tools
headroom_compress/headroom_retrieve - AST-aware code compressors (Python, JS, Go, Rust, Java, C++)
- Cross-agent memory (Claude, Codex, Gemini) with auto-dedup
headroom learnmines failed sessions and writes corrections
Install
As an MCP server (works with Claude, Codex, Cursor, Aider, Copilot). See the repo for the exact config; typical Python install:
pip install headroom
Then register the MCP server in your client, or run the proxy / use the library directly.
- Repo: https://github.com/chopratejas/headroom
- License: Apache-2.0