Headroom
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. Same answers, 60 to 95 percent fewer tokens.
Headroom is a context compression layer for AI agents. It compresses everything your agent reads, including tool outputs, logs, RAG chunks, files and conversation history, before any of it reaches the model. The answers stay the same at a fraction of the tokens.
You can use it three ways: as a library with a single compress() call in Python or TypeScript, as a drop-in proxy that needs no code changes in any language, or as an MCP server. On typical agent workloads it reports 60 to 95 percent fewer tokens.
2 Reviews
Log in to leave a review.
Shipped this into my side project the same night I found it. Token bill dropped by more than half. Answers look identical. That is honestly the whole review. One compress call in TypeScript and my agent stopped eating my API credits for breakfast. If you are a solo dev watching your usage dashboard with a knot in your stomach, just try it.
I plugged Headroom in front of a retrieval pipeline that was quietly bankrupting us on tool-output tokens. The proxy mode meant I changed precisely zero lines of application code, which is the only reason I tried it on a Friday afternoon instead of scheduling it for never.
The compression held up on the part I cared about, which is answer quality. I ran our eval set before and after and the scores moved within noise while token counts dropped by roughly seventy percent on the log-heavy traces. Big tool dumps and RAG chunks are where it earns its keep, because that is where my context was mostly redundant boilerplate anyway.
It is less dramatic on already-tight prompts, which is fair, there is nothing to squeeze. I would like a bit more visibility into what it dropped, ideally a diff I can audit. But for the money this saves on a busy agent, I am not going to complain loudly.