MCP Chinese Getting Started Guide

🚀 Model Context Protocol (MCP) Getting Started Tutorial

This guide provides detailed instructions on how to quickly get started with the Model Context Protocol (MCP) and deploy it to the cloud. It covers the entire process from installing dependencies, creating an MCP service, to final cloud deployment, along with relevant code examples.

🚀 Quick Start

📦 Installation

Before you begin, make sure you have installed the following dependencies:

• Python 3.x
• pip (for managing Python packages)
• Node.js (optional, only for front - end development)

Run the following command in your command - line tool to install the required Python packages:

pip install mcp

✨ Features - Creating an MCP Service

Basic Usage

Assume you have encapsulated your model according to the official documentation or example code. Then, you can use the following code to create a simple MCP service:

# Example code for a Model Context Protocol service

from mcp.server import MCPModelService, Context

class MyModelService(MCPModelService):
    def __init__(self, context: Context):
        super().__init__(context)
        # Initialize the model or related components

    async def __call__(self, inputs):
        """Input processing function"""
        pass

# Create a service instance
if __name__ == '__main__':
    service = MyModelService(Context())
    service.run()

Runtime Configuration

Create an mcp.json file in the root directory of your project to define the runtime configuration of the MCP service:

{
  "model": {
    "name": "my-model",
    "version": "1.0.0",
    "description": "This is a simple example of a Model Context Protocol service.",
    "author": "Your name or company name"
  },
  "server": {
    "port": 5000,
    "workers": 4
  }
}

🔧 Deployment to the Cloud

Selecting a Cloud Platform

Currently, MCP supports deployment on multiple cloud platforms. This guide uses Alibaba Cloud Function Compute (FC) as an example.

Creating a Function Compute Service

Log in to the Alibaba Cloud console, go to the Function Compute service page, and follow these steps to create a new function:

1. Click "Create Function".
2. Fill in the function name, select the runtime environment (it is recommended to choose "MCP Runtime"), and complete other configurations.
3. Upload your code and dependencies.

Deploying the MCP Service

In the Function Compute console, go to the function page you just created and click "Deploy".

💻 Usage Examples

Client - side Testing

Use the following Python code to test your MCP service:

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


async def main():
    async with sse_client('https://mcp-test-whhergsbso.cn-hangzhou.fcapp.run/sse') as streams:
        async with ClientSession(*streams) as session:
            await session.initialize()

            res = await session.call_tool('web_search', {'query': 'Test query'})
            print(res)


if __name__ == '__main__':
    asyncio.run(main())

Tool Configuration

You can configure the following tools by adding the context information of the MCP service:

• VS Code: Add debugging configurations in the launch.json file.
• PyCharm: Install the MCP support plugin from the plugin marketplace.
• Docker: Use the pre - built image for quick deployment.

📚 Documentation - Update Log

April 9, 2023

• Added the "MCP Runtime" option, simplifying the process of creating a Function Compute service.
• Fixed some known issues and optimized performance.

For more detailed information, please refer to the MCP official documentation or visit the GitHub repository for support.