ntthanh2603/gemini-web-to-api ? reverse-engineered prompt
Reverse engineered prompt
Build me a small local server that turns the Gemini web app into a REST API I can use like OpenAI, Claude, or Gemini.
I want to sign in with my browser cookies instead of an API key, then send chat requests to a localhost endpoint and get normal AI responses back. Please include easy setup with environment variables, Docker support, and simple API docs I can open in the browser. It should handle session refresh automatically, have basic rate limiting, and a health check so I know it is running.
If possible, make it support image generation and image inputs too, since I may want to send URLs, data URLs, or a few reference images in the same request. Keep it practical and ready to run, with clear examples for testing it from the command line. If you need to look up current docs online, feel free to do that.