Alishahryar1/free-claude-code ? reverse-engineered prompt

I want to build a tool that lets me use the official Claude Code CLI and VSCode extension for free, without needing a paid Anthropic API key.

The idea is to create a lightweight Python proxy server that intercepts the API calls and redirects them to other model providers. It should support free backends like NVIDIA NIM and OpenRouter, but also let me connect to models I'm running locally with LM Studio or llama.cpp.

Configuration should be simple through an .env file. I'd like to be able to map different Claude models like Opus, Sonnet, and Haiku to different providers. For instance, I could use a powerful free model for my "Opus" requests and a fast local one for "Haiku".

The server should be smart about handling API rate limits for me. As a bonus, please add a simple Discord bot integration so I can run coding tasks from my server. Use Uvicorn to run the server.