PromptEngineer48/free-claude-code ? reverse-engineered prompt

Build me a small Python app that lets me use Claude Code through my own proxy so I can point the Claude CLI or VS Code at it and use free or local models instead of Anthropic directly.

I want one simple service I can run locally, with an easy config file, that can forward Claude style requests to a few different backends like NVIDIA NIM, OpenRouter, DeepSeek, LM Studio, llama.cpp, or Ollama. It should let me choose a default model and also route different model tiers to different providers when needed. Please make it work with Claude Code’s model picker, streaming, tool use, and the usual chat flow.

If it makes sense, add a small bot mode for Discord or Telegram, and voice note transcription for remote coding chats. Keep the setup straightforward, include a good example env file, and make sure I can start it quickly and connect Claude Code to it with minimal hassle.