Skip to content

README.md

Latest commit

 

History

History
237 lines (185 loc) · 7.12 KB

README.md

File metadata and controls

237 lines (185 loc) · 7.12 KB

claude-max-api-proxy-rs

CI

Use your Claude Max subscription with any OpenAI or Anthropic-compatible client — no separate API costs.

A fast Rust proxy that wraps the Claude Code CLI as a subprocess and exposes both OpenAI and Anthropic HTTP APIs. Any tool that speaks either protocol can now use your Max subscription directly.

Why

Approach Cost Limitation
Anthropic API ~$15/M input, ~$75/M output tokens Pay per use
Claude Max $200/month flat OAuth blocked for third-party API use
This proxy $0 extra (uses Max subscription) Routes through CLI

Anthropic blocks OAuth tokens from third-party API clients, but the Claude Code CLI can use them. This proxy bridges that gap.

How It Works

Your App (OpenAI or Anthropic client)
         │
         ▼
    HTTP Request (either format)
         │
         ▼
   claude-max-api-proxy-rs
         │
         ▼
   Claude Code CLI (subprocess)
         │
         ▼
   Anthropic API (via Max OAuth)
         │
         ▼
   Response → Your format → Your App

Features

  • Dual API support — OpenAI /v1/chat/completions and Anthropic /v1/messages on the same server
  • Streaming — Real-time SSE for both protocols, matching their native event formats
  • Model mapping — Flexible name resolution (claude-opus-4, claude-sonnet-4-20250514, opus, etc.)
  • Session management — Conversation continuity via persistent session IDs
  • Zero config — Uses existing Claude CLI auth, no API keys to manage
  • Fast — Native Rust binary, ~3MB stripped. Starts instantly.
  • Safe — No shell execution; all subprocess args passed directly

Prerequisites

  1. Claude Max subscription — Subscribe here
  2. Claude Code CLI installed and authenticated: 
    npm install -g @anthropic-ai/claude-code
claude auth login
  3. Rust toolchain (for building from source): 
    curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

Installation

git clone https://github.com/thhuang/claude-max-api-proxy-rs.git
cd claude-max-api-proxy-rs
cargo build --release

The binary is at target/release/claude-max-api.

Usage

# Default (port 8080)
claude-max-api

# Custom port
claude-max-api 3456

# Custom port + working directory for CLI subprocesses
claude-max-api 8080 --cwd ~/projects

The server binds to 127.0.0.1 (localhost only).

Quick test

# Health check
curl http://localhost:8080/health

# List models
curl http://localhost:8080/v1/models

# OpenAI format
curl -X POST http://localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-opus-4",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

# Anthropic format
curl -X POST http://localhost:8080/v1/messages \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-opus-4",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

# Streaming (add -N to disable curl buffering)
curl -N -X POST http://localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4",
    "messages": [{"role": "user", "content": "Hello!"}],
    "stream": true
  }'

API

Endpoint Method Description
/health GET Health check with uptime
/v1/models GET OpenAI-compatible model list
/v1/chat/completions POST OpenAI Chat Completions (streaming & non-streaming)
/v1/messages POST Anthropic Messages (streaming & non-streaming)

Models

Model ID CLI Alias Context Window Max Output
claude-opus-4 opus 1,000,000 128,000
claude-sonnet-4 sonnet 200,000 64,000
claude-haiku-4 haiku 200,000 64,000

Date-suffixed variants (e.g. claude-opus-4-20250514) and claude-code-cli/ prefixed names are also accepted.

Client Examples

Python (OpenAI SDK)

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:8080/v1",
    api_key="not-needed"
)

response = client.chat.completions.create(
    model="claude-opus-4",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)

Python (Anthropic SDK)

import anthropic

client = anthropic.Anthropic(
    base_url="http://localhost:8080",
    api_key="not-needed"
)

message = client.messages.create(
    model="claude-opus-4",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello!"}]
)
print(message.content[0].text)

Continue.dev

{
  "models": [{
    "title": "Claude (Max)",
    "provider": "openai",
    "model": "claude-opus-4",
    "apiBase": "http://localhost:8080/v1",
    "apiKey": "not-needed"
  }]
}

Architecture

src/
├── main.rs           # CLI args, startup checks, graceful shutdown
├── server.rs         # Axum router, CORS, middleware
├── routes.rs         # Endpoint handlers (health, models, completions, messages)
├── subprocess.rs     # Claude CLI process lifecycle and NDJSON parsing
├── session.rs        # Session persistence (~/.claude-code-cli-sessions.json)
├── error.rs          # Unified error types → HTTP responses
├── types/
│   ├── openai.rs     # OpenAI request/response types
│   ├── anthropic.rs  # Anthropic request/response types
│   └── claude_cli.rs # CLI NDJSON message types
└── adapter/
    ├── openai_to_cli.rs    # OpenAI request → CLI invocation
    ├── cli_to_openai.rs    # CLI output → OpenAI response
    ├── anthropic_to_cli.rs # Anthropic request → CLI invocation
    └── cli_to_anthropic.rs # CLI output → Anthropic response

Environment Variables

Variable Default Description
RUST_LOG claude_max_api=info Log level filter (tracing env-filter syntax)

Acknowledgements

This project is inspired by claude-max-api-proxy, the Node.js proxy documented by the OpenClaw project. That work demonstrated the core idea — wrapping the Claude Code CLI as a local API server to unlock Max subscription access for third-party clients. This Rust rewrite builds on that foundation with native Anthropic API support and a leaner runtime.

Comparison with Node.js Version

Node.js Rust
API support OpenAI only OpenAI + Anthropic
Binary size ~200MB (with node_modules) ~3MB
Startup ~500ms <10ms
Dependencies Express + many transitive deps Axum + minimal deps
Build npm install && npm run build cargo build --release

License

MIT