pwnapplehat/cursor-openai

cursor-openai-gateway lets you use your existing Cursor editor subscription as if it were an OpenAI API endpoint. If you have a tool, library, or script that expects to talk to OpenAI's API, you can point it at this local server instead, and your requests will actually run through a real Cursor AI agent using your own Cursor account, giving you access to more than eighty models without paying for a separate API key. The project ships with a web dashboard that walks you through setup the first time you run it, then gives you a full control panel afterward: live activity monitoring, session management, a browsable list of available models, and a real chat window for testing. Everything the dashboard can do is also available from a dedicated command line tool for scripting, and every setting can be configured entirely through environment variables for running it inside Docker or on a server with no interface at all. It supports streaming responses, multi-turn conversations that remember earlier messages in a session, and OpenAI style tool calling, where the AI can request that your own code run a function and return the result. To get started you need Node.js version 22.13 or newer. On Windows you double click a provided batch file, and on Mac or Linux you run a shell script, either way, the script installs everything needed, builds the project, starts the server, and opens the setup wizard in your browser automatically, where you paste in a Cursor API key and pick a default model. The maintainer specifically tested this project on Windows throughout development and separately verified it end to end on Linux with a fresh clone, covering the full test suite, the dashboard, streaming chat, and the command line tool. macOS support is written to mirror the Linux code paths but has not been physically tested on a Mac.