MCP_Server

A Model Context Protocol (MCP) Server implementation for training purposes.

Overview

This project demonstrates how to create and integrate an MCP Server with various components including:

• RAG (Retrieval-Augmented Generation) implementation
• LangChain integration
• Ollama model support
• Data preprocessing and embedding storage

Project Structure

MCP_Server/
├── MCP/
│   ├── load_data.py          # Data loading utilities
│   ├── preprocess.py         # Data preprocessing pipeline
│   ├── embedding_store.py    # Vector embedding storage and retrieval
│   ├── rag_query.py          # RAG query interface
│   └── main.py               # Main entry point
├── MCP_Server/
│   ├── dummy_mcp_server.py   # MCP Server implementation
│   └── test_client.py        # Test client for MCP Server
├── Notebooks/
│   ├── ollama_course.ipynb   # Ollama course materials
│   └── langchain_retrieval.ipynb
├── assets/
│   ├── filtered_dataset.csv  # Training data
│   └── retrieval_resources/
└── README.md

Getting Started

Prerequisites

• Python 3.8+
• Ollama (for model inference)
• LangChain
• langchain-mcp-adapters
• Required dependencies listed below

Installation

1. Clone the repository:

git clone <repository-url>
cd MCP_Server

2. Install dependencies:

pip install langchain-ollama langchain-mcp-adapters langchain

3. Set up Ollama and pull a model:

ollama pull llama3.2:1b

Usage

Running the MCP Server

Start the MCP server:

python MCP_Server/dummy_mcp_server.py

Testing with the Client

Run the test client to verify the server:

python MCP_Server/test_client.py

This will:

• Load available tools from the MCP server
• Demonstrate LLM agent integration with tools
• Show direct tool invocation

Running RAG Queries

Use the RAG implementation to perform retrieval-augmented generation:

python MCP/rag_query.py

Perform retrieval-augmented generation using embeddings

Uses custom National Disaster dataset for demonstration

ℹ️ Note: The National Disaster dataset used in this project is included in assets/filtered_dataset.csv. You can access and explore it directly for experimentation.

Available Tools

The MCP server provides the following tools for LLM integration:

Utility Tools

• random_number(min_val: int = 0, max_val: int = 100) -> int: Generates a random integer between specified bounds
• reverse_string(text: str) -> str: Reverses the input string
• is_even(n: int) -> bool: Checks if a number is even
• get_country_capital(country: str) -> str: Returns the capital of a given country (supports Switzerland, France, Germany)

Mathematical Tools

• calculate(expression: str) -> float: Evaluates simple mathematical expressions safely
• list_operations(numbers: list[int]) -> dict: Returns statistics (sum, average, min, max, count) for a list of numbers

Text Processing Tools

• word_count(text: str) -> dict: Returns word count statistics including total words, characters, and average word length

Conversion Tools

• convert_temperature(temp: float, from_unit: str, to_unit: str) -> float: Converts temperature between Celsius, Fahrenheit, and Kelvin

Date/Time Tools

• get_current_datetime() -> str: Returns the current date and time as a formatted string

Key Features

• MCP Server Implementation: Stdio-based transport for tool communication
• Comprehensive Tool Set: 9 different tools covering various utility functions
• LangChain Agent: LLM-powered agent using ChatOllama
• RAG Support: Retrieval-augmented generation with embeddings
• Data Pipeline: Complete data loading and preprocessing workflow

Troubleshooting

Import Error: langchain_mcp_adapters

Ensure the package is installed:

pip install langchain-mcp-adapters

Restart the Python interpreter in VS Code:

• Press Ctrl+Shift+P
• Search for "Python: Restart Language Server"

License