PublicAffairs/openai-gemini ? reverse-engineered prompt
Reverse engineered prompt
Build me a small serverless proxy that lets apps using the OpenAI API talk to Gemini instead.
I want to give it my Google API key, deploy it easily to places like Vercel, Netlify, Cloudflare, or run it locally, and then point my tools at the new base URL instead of OpenAI. It should feel like an OpenAI compatible endpoint, with chat completions, embeddings, and model listing working well enough for common apps. Support streaming too, plus image and audio inputs when the request format allows it. If possible, let me pass through Gemini specific options for things like thinking or search, but keep the normal OpenAI style request and response shape as close as possible.
If you need to check current platform docs while wiring it up, go ahead. Make sure the setup is simple and the README style instructions are clear enough for someone who just wants it working fast.
Are you gonna build this?
make sure you review the code using coderabbit