wtf is cursor-plan2api?

Cursor-Plan2API is a small local program that lets people use their existing Cursor.ai subscription as if it were a regular API service, similar to the ones offered by OpenAI. Normally, using Cursor means working inside its own editor or command line tool. This project runs a local server on your own computer that translates requests written in the common OpenAI format into calls Cursor's own agent can understand, and then sends the replies back in the same format other tools expect. Because it speaks the same language as OpenAI's API, other programs that already know how to talk to OpenAI, such as Hermes Agent, OpenCode, the OpenAI Python and JavaScript libraries, LangChain, and automation tools like n8n, can be pointed at this local server instead. It also understands the format Claude Code uses, and can generate images and text embeddings. The tool supports both streaming and one shot responses, so replies can appear word by word or all at once. It can hold onto conversation state so multi-turn conversations do not need to resend the whole history each time, and it can compress long conversations to keep them running smoothly. There is an optional pool of ready CLI connections to speed up repeated requests, and requests that fail because of rate limits are retried automatically. It ships with a Docker setup for people who prefer containers, a simple web dashboard to watch live usage statistics, and a small built in web page for trying it out without writing any code. Setup mainly involves installing the Cursor command line tool, logging in with an existing Cursor.ai account, then installing this project through npm or from source. The project is written in TypeScript, runs on Node.js version 20 or newer, and is released under the MIT license, so it can be freely used, changed, and shared as long as the copyright notice stays intact. It currently has 4 stars on GitHub.