# llama-connect

Allow web UI to connect to a locally-hosted `llama-server` instance via WebRTC peer-to-peer connection.

Frontend protocol: https://github.com/ggml-org/llama.cpp/pull/24577

## Demo

```sh
# for example, llama-server is running on port 8080

llama-connect --port 8080
```

Then open [client.html](examples/client.html) demo to connect to it. The chat completion request will be proxied to local `llama-server` via WebRTC connection.