claudex

A CLI proxy to run Claude API requests (Anthropic-style) against OpenAI-compatible LLM providers (like OpenRouter), enabling Claude Code to work with any OpenAI-compatible endpoint.

Requirements

Python 3.8+
uvicorn for ASGI server
FastAPI, httpx, python-dotenv, pydantic (see pyproject.toml)

Setup

git clone https://github.com/nenadilic84/claudex.git
cd claudex
python3 -m venv .venv
source .venv/bin/activate
pip install -e .
cp .env.example .env  # edit .env to fill in your API settings

Fill in your .env like:

TARGET_API_BASE=https://api.openrouter.ai/v1
TARGET_API_KEY=<your_provider_key>
BIG_MODEL_TARGET=openai/gpt-4.1
SMALL_MODEL_TARGET=openai/gpt-4.1-mini
LOG_LEVEL=INFO

Usage

After setup and installing dependencies, you can run the proxy in either of these ways:

1. Recommended: Run via the CLI/main entrypoint

# Run as module with environment variables:
python -m claudex.main --host 0.0.0.0 --port 8082 --reload

# Or with command line arguments:
python -m claudex.main --target-api-base https://api.openrouter.ai/v1 --big-model openai/gpt-4.1 --small-model openai/gpt-4.1-mini --log-level INFO

# Or if installed as a script:
claudex --host 0.0.0.0 --port 8082 --target-api-base https://api.openrouter.ai/v1 --big-model openai/gpt-4.1 --small-model openai/gpt-4.1-mini

CLI Options

--host - Host to bind (default: 0.0.0.0)
--port - Port to bind (default: 8082)
--reload - Enable live-reload for development
--target-api-base - Base URL for target OpenAI-compatible API
--big-model - Target model for Claude Sonnet/larger models
--small-model - Target model for Claude Haiku/smaller models
--log-level - Logging level (DEBUG, INFO, WARNING, ERROR)

Note: The API key must still be provided via the TARGET_API_KEY environment variable for security reasons.

2. Alternative: Run directly with Uvicorn

uvicorn claudex.proxy:app --host 0.0.0.0 --port 8082 --reload

Using with Claude CLI

In a second terminal, you can now use the Claude CLI tool with this proxy:

ANTHROPIC_BASE_URL=http://localhost:8082 DISABLE_PROMPT_CACHING=1 claude

This allows you to use Claude Code with any OpenAI-compatible LLM provider, such as:

OpenRouter
Together.ai
Local LLM endpoints
Any other OpenAI-compatible API

API Endpoints

GET / - Health check endpoint
POST /v1/messages - Main endpoint that receives Anthropic API requests, converts them to OpenAI format, and returns converted responses

Development

Run tests:

pytest

2.6 KiB Raw Permalink Blame History