CC-Proxy-Executable 🔄

Use Anthropic clients (like Claude Code) with any OpenAI-compatible API, including Gemini, OpenAI, or custom endpoints. 🤝

A proxy server that lets you use Anthropic clients with Gemini, OpenAI, or any other OpenAI-compatible API, all via LiteLLM. 🌉

Quick Start ⚡

The only prerequisite is to have uv installed.

curl -LsSf https://astral.sh/uv/install.sh | sh

Installation and Usage

You can install and run the proxy directly from GitHub with a single command. You will need to provide an API key for the service you want to use.

For OpenAI, you would run:

LLM_API_KEY="your-openai-api-key" uvx https://github.com/minpeter/cc-proxy-executable.git

This will start the proxy server on http://localhost:8082.

Using with Claude Code 🎮

1. Install Claude Code (if you haven't already):

   npm install -g @anthropic-ai/claude-code

2. Connect to your proxy:

   ANTHROPIC_BASE_URL=http://localhost:8082 claude

3. That's it! Your Claude Code client will now use the configured backend models through the proxy. 🎯

Using Custom OpenAI-Compatible APIs

You can easily switch to any OpenAI-compatible API by setting the LLM_BASE_URL and LLM_API_KEY environment variables.

Example: Using Friendli.ai

To use the proxy with friendli.ai, you would run the following command:

LLM_BASE_URL="https://api.friendli.ai/serverless/v1" \
LLM_API_KEY="your-friendli-token" \
BIG_MODEL="Qwen/Qwen3-32B" \
SMALL_MODEL="Qwen/Qwen3-235B-A22B-Instruct-2507" \
uvx https://github.com/minpeter/cc-proxy-executable.git

In this example:

• LLM_BASE_URL is set to the Friendli.ai serverless endpoint.
• LLM_API_KEY should be your FRIENDLI_TOKEN.
• BIG_MODEL and SMALL_MODEL are set to the desired model on Friendli.ai.

How It Works 🧩

This proxy works by:

1. Receiving requests in Anthropic's API format 📥
2. Translating the requests to OpenAI format via LiteLLM 🔄
3. Sending the translated request to the configured backend 📤
4. Converting the response back to Anthropic format 🔄
5. Returning the formatted response to the client ✅

The proxy handles both streaming and non-streaming responses, maintaining compatibility with all Claude clients. 🌊