Container MCP

🚀 Containerized Task Platform Documentation

This is a containerized task execution platform (MCP) designed to provide users with a secure, isolated, and scalable task running environment. The platform uses Docker container technology to achieve task isolation and resource control, and provides real-time task execution status updates through the Server-Sent Events (SSE) protocol.

✨ Features

• Task Isolation: Each task runs in an independent Docker container, ensuring that tasks do not interfere with each other.
• Resource Limitation: Supports hard limits on CPU, memory, and disk space to prevent resource abuse.
• Task Monitoring: Provides real-time task execution status updates and allows users to terminate abnormal tasks.
• Multi-Tenant Support: Resource isolation and permission control for different users can be achieved through configuration.

📦 Installation

Quick Installation

1. Clone the project repository:

   git clone https://github.com/yourusername/container-mcp.git
   cd container-mcp
   

2. Download and install Docker:

   • On Linux systems, use the official script to install:

     curl -fsSL https://get.docker.com | bash -s docker
     

   • On Windows or macOS, visit the Docker official website to download and install.

3. Start and join the DockerGroupName group (Linux only):

   sudo groupadd docker
   sudo usermod -aG docker $USER
   newgrp docker
   

4. Build the project container:

   ./bin/02-build-container.sh
   

5. Configure the running environment:

   ./bin/03-setup-environment.sh
   

6. Start the MCP service:

   ./bin/04-run-container.sh
   

Detailed Installation Steps

1. Initialize the project:

   ./bin/01-init.sh
   

2. Build the container image:

   ./bin/02-build-container.sh
   

3. Configure the running environment:

   ./bin/03-setup-environment.sh
   

4. Start the MCP service:

   ./bin/04-run-container.sh
   

5. Run tests (optional):

   ./bin/05-run-tests.sh
   

💻 Usage Examples

Basic Usage

After the MCP service is started, any MCP client can connect to the service. By default, the service runs at http://localhost:8000 or the address specified in the configuration file.

⚠️ Important Note

When configuring the MCP client, the target address must be set to http://127.0.0.1:<port>/sse (where <port> is the default port 8000 or other configured ports). The path /sse is required for server-sent event communication.

Example Python Client

from mcp.client.sse import sse_client
from mcp import ClientSession
import asyncio

async def main():
    # Connect to the MCP service
    sse_url = "http://127.0.0.1:8000/sse"  # Or other configured ports
    
    # Initialize the SSE client connection
    async with sse_client(sse_url) as client:
        # Create an MCP session
        session = ClientSession(client)
        
        # Get task execution status updates
        while True:
            update = await session.receive_update()
            print(f"Received update: {update}")

asyncio.run(main())

📚 Documentation

Configuration Guide

MCP supports configuration through environment variables. The following are common configuration items:

Example Configuration

Before running the MCP service, you can adjust its behavior by setting environment variables:

# Enable debug mode and specify the image registry
export MCP_DEBUG=true
export DOCKER_REGISTRY=private-registry:8080
./bin/04-run-container.sh

Development Guide

Environment Preparation

1. Install Python 3.8 or higher:

   • On Linux, use the package manager to install:

     sudo apt-get install python3 python3-pip
     

   • On Windows and macOS, visit the Python official website to download and install.

2. Install project dependencies:

   pip install -r requirements.txt
   

Code Structure

The project code mainly consists of the following parts:

• container_mcp/: Contains the core implementation of containerized tasks.
• mcp_client/: Provides a client interface for interacting with the MCP service.
• tasks/: Stores executable task scripts or binary files.

Development Process

1. Clone the project repository:

   git clone https://github.com/yourusername/container-mcp.git
   cd container-mcp
   

2. Install dependencies:

   pip install -r requirements.txt
   

3. Start the development server:

   python -m mcp.start_development_server
   

📄 License

This project is licensed under the MIT License. Please refer to the LICENSE.md file for specific license details.

Author Information

• Developer: Your Name
• Contact Email: your.email@example.com
• GitHub Repository: Container-MCP

Thank you for using Container-MCP! If you need help or have feedback, please feel free to contact us.