Model Context Protocol Server and Tools

This repository contains implementations of Model Context Protocol (MCP) servers and client applications that demonstrate how to use MCP for AI agent interactions.

Part 1: Math Tools Integration (client.py and mathserver.py)

This part of the codebase demonstrates how to create a simple MCP server that provides math operations and a client that utilizes these operations through a language model.

mathserver.py

A simple MCP server that provides basic mathematical operations:

• add: Adds two integers together
• subtract: Subtracts the second integer from the first
• multiply: Multiplies two integers together
• divide: Divides the first integer by the second, returns a float

The server uses FastMCP to easily expose these operations as tools that can be called by MCP clients.

client.py

A demonstration client that:

1. Sets up a connection to the math server using MultiServerMCPClient
2. Uses the Groq API with the "qwen-qwq-32b" model
3. Creates a ReAct agent using LangGraph
4. Executes a simple math problem "(88*(2+3))/5" by having the language model use the math tools

How to Run

1. Make sure you have all dependencies installed: pip install -r requirements.txt
2. Set your Groq API key in a .env file: GROQ_API_KEY=your-api-key
3. Run the client: python client.py

The client will automatically start the math server as a subprocess, communicate with it using stdio transport, and solve the math problem.

Part 2: Advanced MCP Integration (app.py and mcpconfig.json)

This part demonstrates a more advanced usage of MCP with multiple external tool servers and an interactive chat interface.

mcpconfig.json

Configuration file that defines external MCP servers:

• playwright: For web automation tasks
• airbnb: For Airbnb-related operations
• duckduckgo-search: For web search functionality

Each server is configured with the command and arguments needed to start it.

app.py

An interactive chat application that:

1. Connects to all the servers defined in mcpconfig.json
2. Sets up a chat interface using the Groq "qwen-qwq-32b" model
3. Creates an MCPAgent that can use all available tools from the configured servers
4. Provides a memory-enabled conversation that maintains chat history
5. Allows users to interact with the agent in a continuous conversation

Features:

• Persistent conversation history (memory)
• Commands to exit the chat (exit, quit)
• Command to clear conversation history (clear)
• Error handling for agent operations

How to Run

1. Make sure you have all dependencies installed: pip install -r requirements.txt
2. Set your Groq API key in a .env file: GROQ_API_KEY=your-api-key
3. Run the application: python app.py
4. Interact with the chat interface by typing messages

Note: When running app.py, the external MCP servers will be started automatically as configured in mcpconfig.json.

Requirements

See requirements.txt for the full list of dependencies:

• langchain-groq
• langchain-mcp-adapters
• mcp
• langgraph
• mcp-use