🤖 manusIM-proxy

A lightweight OpenAI-compatible bridge for the Manus IM agent.

English • 中文 • 日本語 • Español • Français

---

manusIM-proxy is a high-performance HTTP bridge that translates OpenAI-style chat completion requests into Manus v2 API operations. It seamlessly converts Manus task events into standard OpenAI JSON or Server-Sent Events (SSE), enabling you to use Manus with your favorite AI tools like 9router, OpenCode, Claude Code, or any OpenAI-compatible SDK.

Important

This project uses the official Manus API. Keep your API keys secure, follow Manus usage terms, and never expose this service to the public internet without proper authentication.

---

✨ Features

• OpenAI Compatibility: Drop-in replacement for OpenAI endpoints (/v1/chat/completions, /v1/models).
• Streaming Support: Real-time task event streaming via SSE.
• Token Pooling: Intelligent round-robin token leasing with failover and cooldown mechanisms.
• Agent Registry: Automatically discovers and maps Manus agents to OpenAI models.
• Container Ready: Optimized Docker and Docker Compose support for easy deployment.

📸 Preview

🚀 Quick Start

Local Installation

1. Clone and Setup

   git clone https://github.com/0xgetz/manusIM-proxy.git
   cd manusIM-proxy
   python3 -m venv .venv
   source .venv/bin/activate
   pip install -r requirements.txt

2. Configure Environment

   cp .env.example .env
   # Add your Manus API keys to AUTH_TOKENS in .env

3. Launch

   python -m app.main

Using Docker

docker compose up -d --build

---

⚙️ Configuration

Variable	Description	Default
AUTH_TOKENS	Comma-separated Manus API keys	Required
PORT	Service listening port	3457
MANUS_API_BASE	Upstream Manus API URL	https://api.manus.ai
REGISTRY_REFRESH	Interval to refresh agent list (e.g., 6h)	6h
SESSION_TIMEOUT	Max polling time for a task (e.g., 30m)	30m
TASK_ROTATION	Max reuse time for a created task	10m
DEFAULT_TASK_ID	Pin agent-default-main_task to a specific ID	Optional

---

🛠️ Usage

Python SDK Example

from openai import OpenAI

client = OpenAI(
    base_url="http://127.0.0.1:3457/v1",
    api_key="proxy-key" # Any string works
)

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

cURL Example

curl http://localhost:3457/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "manus-1.6",
    "messages": [{"role": "user", "content": "Explain the task lifecycle."}],
    "stream": true
  }'

---

🔌 Integration with 9router

1. Go to Providers → Custom Providers → Add OpenAI Compatible.
2. Base URL: http://127.0.0.1:3457/v1
3. Model: agent-default-main_task
4. Click Create and start using Manus through the 9router gateway!

---

🧪 Testing

Run the local test suite to verify functionality:

python3 test_pool.py
python3 test_endpoints.py

---

📚 References

• Manus API Authentication
• Task Lifecycle Documentation
• Agent List API

---

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.