MCP Python Tutorial

🚀 MCP Python Tutorial

A tutorial application for MCP in Python using a simple local database and mock data.

🚀 Quick Start

📦 Installation

Install via Smithery

To automatically install the Python MCP tutorial server for the Claude desktop version via Smithery:

npx -y @smithery/cli install @jhj0517/mcp-python-tutorial --client claude

Manual Installation

1. Clone this repository.
2. Install the dependencies:

pip install -r requirements.txt

3. Run the MCP server in development mode:

mcp dev localdb_app.py

4. The default port of the MCP server is 5173. Visit http://localhost:5173.

✨ Features

This tutorial application demonstrates the core concepts of MCP. You can view the role-based annotations in tutorial_app/mcp_server.py:

@mcp.resource

Basically, this annotation is about the agent "fetching" resources, similar to GET in a REST API.

• users://all - Fetch all users.
• users://{user_id}/profile - Fetch a user's profile.
• posts://all - Fetch all posts.
• posts://{post_id} - Fetch a post by ID.

@mcp.tool

This is about the agent "generating" new resources, similar to POST in a REST API.

• create_user - Create a new user.
• create_post - Create a new post.
• search_posts - Search for posts by title or content.

@mcp.prompt

This is just a reusable template for convenient interaction with the LLM.

• user_profile_analysis - Generate a user profile analysis.
• post_feedback - An interactive prompt for post feedback.

⚠️ Important Note

For more annotations, read: https://github.com/modelcontextprotocol/python-sdk?tab=readme-ov-file#核心概念

🔗 Connecting to the Client

Once you've set up the MCP server, you need an LLM client to build an agent using your MCP server. The following guide will help you use Claude Desktop as the client.

1. Claude Desktop uses uv to install the MCP server dependencies. First, install uv:

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

2. Install the MCP server dependencies using uv:

# Create a virtual environment and activate it
uv venv
.venv\Scripts\activate

uv pip install -r requirements.txt

3. Download Claude Desktop from:

• Link

4. Find or create the claude_desktop_config.json file. The location varies depending on the operating system:

• Windows:

C:\Users\%USER%\AppData\Roaming\Claude\claude_desktop_config.json

• MacOS/Linux:

~/Library/Application\ Support/Claude/claude_desktop_config.json

5. Add the mcpServers property to claude_desktop_config.json:

{
    "mcpServers": {
        "local_db": {
            "command": "uv",
            "args": [
                "--directory",
                "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather",
                "run",
                "localdb_app.py"
            ]
        }
    }
}

Note: You can deploy multiple MCP servers, each with its own responsibilities and areas of expertise. This separation of concerns is better than implementing everything in a single MCP server.