Status: [UNTESTED] Endpoints found: http://localhost:8080/chat Source: https://github.com/callmerussell04/gpt-proxy # GPT Proxy Two small Go programs: - `proxy-server`: HTTP proxy to Gemini API. - `proxy-client`: interactive terminal chat client. The server keeps `GEMINI_API_KEY`. The client sends only `GPT_PROXY_TOKEN` to the proxy. ## Server Build a Linux binary for the remote server: ```sh cd proxy-server CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -o proxy-server . ``` Copy it to the server: ```sh scp proxy-server user@your-server:/opt/gpt-proxy/proxy-server ``` Run it on the server: ```sh GEMINI_API_KEY='your-gemini-api-key' \ GPT_PROXY_TOKEN='your-proxy-password' \ GEMINI_MODEL='gemini-2.5-flash-lite' \ /opt/gpt-proxy/proxy-server ``` Optional env: - `ADDR=:8080` - `GEMINI_MODEL=gemini-2.5-flash-lite` - `GEMINI_BASE_URL=https://generativelanguage.googleapis.com/v1beta` - `MAX_CHAT_REQUEST_BYTES=33554432` default request limit for `/chat` bodies, 32 MiB. Create the Gemini API key in Google AI Studio. Free Tier availability and limits depend on the selected model and your Google account. Example with the default Free Tier-oriented model: ```sh GEMINI_API_KEY='your-gemini-api-key' \ GPT_PROXY_TOKEN='your-proxy-password' \ /opt/gpt-proxy/proxy-server ``` Health check does not require auth: ```sh curl http://localhost:8080/health ``` Chat requests require auth: ```sh curl -X POST http://localhost:8080/chat \ -H "Authorization: Bearer $GPT_PROXY_TOKEN" \ -H "Content-Type: application/json" \ -d '{"messages":[{"role":"user","content":"Hello"}]}' ``` For a public server, put HTTPS in front of this service with a reverse proxy. ## Client Docker Build the client image locally: ```sh docker build -t gpt-proxy-client ./proxy-client ``` Run the interactive client: ```sh docker run --rm -it \ -e GPT_PROXY_URL='https://your-server.example.com' \ -e GPT_PROXY_TOKEN='your-proxy-password' \ -v gpt-proxy-client-data:/data \ gpt-proxy-client ``` If the server is exposed directly without HTTPS during local testing: ```sh docker run --rm -it \ -e GPT