Most of a Claude Code session's tokens are not your question. They are the logs, file dumps and tool output Claude reads along the way, most of which it never needed. Headroom is a free, open-source tool that compresses that text on your own computer before it reaches Claude, so the same plan limit goes further.
What it does
- Compresses tool outputs, logs, files and retrieved chunks before they reach the model.
- Runs locally: your prompts and files are not sent anywhere to be compressed.
- The repo's own demo: a log dump went from 10,144 tokens to 1,260, and the FATAL line was still found.
- Works as a library, a local proxy, or an MCP server; for Claude Code you use the proxy with one command.
Free, for your email
Get the full setup steps + the copy-paste setup prompt
The rest of this guide (6 more sections, 6 copy-paste prompts) opens right here. The setup prompt, one paste into Claude that sets it all up, comes by email only. Use your real address.
50,000+ follow where these guides come from. No spam, unsubscribe in one click.