OpenAI-Compatible Claude API Proxy

A lightweight Node.js server that provides an OpenAI-compatible API interface for Claude CLI, allowing you to use Claude with any OpenAI-compatible client.

Features

• OpenAI API Compatibility: Drop-in replacement for OpenAI's chat completions API
• Streaming Support: Real-time streaming responses for interactive applications
• No Timeouts: Requests run without artificial time limits
• Simple Setup: No dependencies, pure Node.js implementation
• Stateless Design: Each request is independent, no session management

Limitations

• No Tool/Function Calling: Tool and function calling features are not supported

Prerequisites

• Node.js (v14 or higher)
• Claude CLI installed and configured (Install Claude CLI)

Installation

1. Clone the repository:

git clone https://github.com/p32929/openai-claude-cli-nodejs.git
cd openai-claude-cli-nodejs

2. No npm install needed - this project has zero dependencies!

Configuration

Environment Variables

Create a .env file in the project root (optional):

# Server port (default: 8000)
PORT=8000

# Enable debug logging
DEBUG=true

# Enable file logging
FILE_LOGGING=true

Usage

Start the Server

# Using default port 8000
npm start

# Or specify a custom port
PORT=3000 npm start

# With debug logging
DEBUG=true npm start

The server will start on http://localhost:8000 (or your specified port).

With Cloudflare Tunnel (for public access)

npm run tunnel

This will start the server and create a public URL using Cloudflare Tunnel.

API Endpoints

Chat Completions

Endpoint: POST /v1/chat/completions

Send messages to Claude and receive responses in OpenAI format.

curl http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "any",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello, how are you?"}
    ],
    "stream": false
  }'

Streaming Example

curl http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -N \
  -d '{
    "model": "any",
    "messages": [
      {"role": "user", "content": "Tell me a story"}
    ],
    "stream": true
  }'

Tool/Function Calling

Tool and function calling is not supported. If you include tools or functions in your request, you'll receive an error:

curl http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "any",
    "messages": [
      {"role": "user", "content": "What is the weather in Paris?"}
    ],
    "tools": [...]
  }'

Error response:

{
  "error": {
    "message": "Tool/function calling is not supported",
    "type": "invalid_request_error"
  }
}

Models List

Endpoint: GET /v1/models

Returns available models (mock response since Claude CLI doesn't enumerate models):

curl http://localhost:8000/v1/models

Health Check

Endpoint: GET /health

Check if the server is running:

curl http://localhost:8000/health

Supported Parameters

Chat Completion Parameters

• messages (required): Array of message objects with role and content
• model: Model name (ignored, Claude CLI uses its default)
• stream: Boolean for streaming responses
• max_tokens: Maximum tokens in response
• temperature: Sampling temperature (0-1)
• top_p: Nucleus sampling parameter
• stop: Stop sequences (string or array)

Compatible with OpenAI Clients

This API works with any OpenAI-compatible client library:

• OpenAI Python SDK
• OpenAI Node.js SDK
• LangChain
• LlamaIndex
• And many more...

Simply point the base URL to http://localhost:8000/v1 and use any model name (e.g., "any").

Architecture

┌─────────────┐     ┌──────────────┐     ┌────────────┐
│   Client    │────▶│   API Proxy  │────▶│ Claude CLI │
│  (OpenAI    │     │   (Node.js)  │     │            │
│  Compatible)│◀────│              │◀────│            │
└─────────────┘     └──────────────┘     └────────────┘

The proxy:

1. Receives OpenAI-formatted requests
2. Converts them to Claude CLI format
3. Executes Claude CLI
4. Transforms responses back to OpenAI format
5. Handles streaming, tool calls, and errors

Disclaimer

This is an unofficial proxy and is not affiliated with Anthropic or OpenAI. Use responsibly and in accordance with Claude's terms of service.