MCP Server Client Tutorial Using Python

🚀 Step-by-step Guide to Building Servers and Clients Using the Model Context Protocol (MCP)

This guide provides a comprehensive walkthrough on constructing servers and clients with the Model Context Protocol (MCP). It details the project structure, installation steps, code implementation, and system testing.

📦 Project Structure

Server Side (server Directory)

• Location: server/server_project_name/server.py
• Content: Contains the core server logic, tool definitions, and component configurations.
• Sample Code:

from mcp_server import Server, Tool

class CalculatorTool(Tool):
    def add(self, a: int, b: int) -> int:
        return a + b

if __name__ == "__main__":
    server = Server()
    server.add_tool(CalculatorTool())
    server.start()

Client Side (client Directory)

• Location: client/mcp_client/server_connector.py
• Content: Includes the logic for connecting to the server and interacting with it.
• Sample Code:

from mcp_client import ServerConnector

connector = ServerConnector("localhost", 1234)
tools = await connector.connect()
calculator_tool = tools["Calculator"]
result = await calculator_tool.add(5, 3)  # Returns 8

🛠️ Steps to Create an MCP Server

Step 1: Install Required Libraries

pip install mcp-server

Step 2: Initialize the Project Structure

• Create the server Directory

mkdir server
cd server

• Initialize Server Code

nano server.py

Add the following content to the file:

from mcp_server import Server, Tool

class CalculatorTool(Tool):
    def add(self, a: int, b: int) -> int:
        return a + b

if __name__ == "__main__":
    server = Server()
    server.add_tool(CalculatorTool())
    server.start()

• Save and Close the File

Step 3: Start the Server

cd ..
python -m server.server_project_name.server

🛠️ Steps to Create an MCP Client

Step 1: Install Required Libraries

pip install mcp-client

Step 2: Initialize the Project Structure

• Create the client Directory

mkdir client
cd client

• Initialize Client Code

nano server_connector.py

Add the following content to the file:

from mcp_client import ServerConnector

async def main():
    connector = ServerConnector("localhost", 1234)
    tools = await connector.connect()
    calculator_tool = tools["Calculator"]
    result = await calculator_tool.add(5, 3)  # Returns 8
    print(f"Result: {result}")

if __name__ == "__main__":
    import asyncio
    asyncio.run(main())

• Save and Close the File

Step 3: Run the Client

cd ..
python -m client.server_connector

🎨 Steps to Create an MCP Client Graphical User Interface (GUI)

Step 1: Install GUI Libraries

pip install tkinter mcp-client

Step 2: Initialize the Project Structure

• Create the client_ui Directory

mkdir client_ui
cd client_ui

• Initialize GUI Code

nano app.py

Add the following content to the file:

import tkinter as tk
from mcp_client import ServerConnector

class CalculatorApp:
    def __init__(self):
        self.window = tk.Tk()
        self.window.title("MCP Calculator Client")
        
        self.entry1 = tk.Entry(self.window)
        self.entry2 = tk.Entry(self.window)
        self.button = tk.Button(self.window, text="Calculate", command=self.calculate)
        self.result_label = tk.Label(self.window, text="")
        
        self.entry1.pack()
        self.entry2.pack()
        self.button.pack()
        self.result_label.pack()
    
    def calculate(self):
        try:
            a = int(self.entry1.get())
            b = int(self.entry2.get())
            connector = ServerConnector("localhost", 1234)
            tools = connector.connect_synchronously()
            calculator_tool = tools["Calculator"]
            result = calculator_tool.add(a, b)
            self.result_label.config(text=f"Result: {result}")
        except Exception as e:
            self.result_label.config(text=f"Error: {str(e)}")

if __name__ == "__main__":
    app = CalculatorApp()
    app.window.mainloop()

• Save and Close the File

Step 3: Run the GUI Application

cd ..
python -m client_ui.app

⚙️ System Expansion

• Add More Tools: Define new tool classes in the server and register them in server.py.
• Enhance the GUI: Add more functions to the client GUI, such as subtraction, multiplication, and division operations.
• Configuration Management: Create configuration files to manage the server address and other settings.
• Logging: Implement logging to track interactions between the client and the server.

🧪 System Testing

Server-side Testing

cd server
python server.py

Client Command-line Testing

cd ..
cd client
python server_connector.py

Client GUI Testing

cd ..
cd client_ui
python app.py

🌟 Conclusion

The Model Context Protocol (MCP) offers a powerful approach to building distributed computing systems. By following this guide, you'll be able to create fully functional servers and clients and expand them to meet various requirements. Continue exploring the potential of MCP and apply it to your projects!