pxpipe
by teamchong
pxpipe is a local proxy that sits between a coding agent (such as Claude Code or Gemini CLI) and the model provider. It turns bulky text context, like system prompts, tool docs and old history, into compact images to reduce input token costs.
- Stars
- 7.6k
- Licence
- MIT
- Runs on
- Your computer
- Setup
- Moderate
- Language
- TypeScript
- Age
- 162 days
What you can do with it
- Run a local proxy that converts bulky agent context into images before it is sent to the model
- Watch tokens saved and each text-to-image conversion on a local dashboard
- Export files, diffs or prompts as PNG pages to upload to image-capable clients such as Cursor
Make money with it
A freelance developer could use it to lower their own AI costs on client projects and so keep more of each fixed-price fee.
An idea, not financial advice. Check the licence before charging for anything built on it.
Ways a business could use it
- Cut the monthly API bill for a small team that uses coding agents heavily
- Fit larger codebases into a single agent session without paying for full text context
Can you use it commercially?
Free to use, change and sell, including in paid products. Keep the copyright notice.
Not legal advice. Read the project's licence file before building a product on it.
Getting it running
Run it with npx (npx pxpipe-proxy), then launch your coding agent through the pxpipe warp command.
Follow the official instructions in the project README.
Things to check
- Newer Claude models (Fable 5.1, Opus 5.5, Sonnet 5.5, Haiku 5.5) read dense images poorly, so savings may come with accuracy loss
- Headline savings figures come from the author's own benchmarks on one older model
- Still needs your own paid API access to the model provider
- Rewrites requests through a local proxy, so check it suits your provider's terms
Who's talking about it
Checked 29 Oct 2026. Summary drafted by AI from the project's README.
