billion-context
by ranxianglei
A proxy that sits between an AI coding agent and its model API and compresses the conversation into layered summaries. It keeps long sessions inside a small context window and cuts the number of tokens you pay for.
Worth a lookEstablished, with a few things to check (listed below).
- Stars
- 3k
- Licence
- MIT
- Runs on
- Your computer
- Setup
- Moderate
- Language
- TypeScript
- Age
- 82 days
What you can do with it
- Run a single AI coding session for days without hitting the context limit
- Cut token costs by having old conversation folded into summaries
- Restore compressed sections or search them when exact details are needed again
Ways a business could use it
- Lower API spend for a freelancer or small team that uses AI coding agents all day
- Keep long-running automated coding or maintenance sessions going on cheaper models with smaller context windows
Can you use it commercially?
Free to use, change and sell (MIT-style terms in the licence file). Some folders may carry other terms.
Not legal advice. Read the project's licence file before building a product on it.
Getting it running
Install globally with npm (npm install -g billion-context --prefix=~/.local), then point your agent's base URL at the proxy or use a launcher such as bili pi.
Follow the official instructions in the project README.
Things to check
- Licence is unclear (not asserted), so check before commercial use
- Needs your own paid model API keys from Anthropic, OpenAI or a compatible provider
- Routes all your prompts and code through a local proxy, so review what it sends and stores
- Compression quality depends on the model deciding what to summarise, so details could be lost
- Documentation is partly in Chinese
Who's talking about it
Checked 27 Oct 2026. Summary drafted by AI from the project's README.
