MCP User Management System

A practical implementation of the Model Context Protocol (MCP) demonstrating a client-server architecture for user management with AI-powered query capabilities.

📖 What is MCP?

The Model Context Protocol (MCP) is an open protocol that standardizes how applications provide context to Large Language Models (LLMs). Think of it as a universal connector between your AI models and various data sources or tools.

Key Concepts:

• MCP Server: Exposes resources (data), tools (actions), and prompts to clients
• MCP Client: Connects to servers and uses their capabilities
• Resources: Read-only data sources (like database queries)
• Tools: Executable functions that can modify state
• Prompts: Reusable prompt templates

This project demonstrates all these concepts through a simple user management system.

🎯 What This Project Does

This repository contains:

1. MCP Server (server.ts): Manages a user database with:

   • Resources to read user data
   • Tools to create users (manually or with AI-generated fake data)
   • Prompts for generating fake user profiles

2. MCP Client (client.ts): Interactive CLI that lets you:

   • Query the system using natural language (AI-powered)
   • Call tools directly
   • Read resources
   • Use prompts
   • Everything integrated with Google's Gemini AI model

🚀 Getting Started

Prerequisites

• Node.js (v18 or higher)
• npm or yarn
• A Google Gemini API key (Get one here)

Installation

1. Clone the repository

   git clone <your-repo-url>
   cd <your-repo-name>
   

2. Install dependencies

   npm install
   

3. Set up environment variables

   Create a .env file in the root directory:

   GEMINI_API_KEY=your_gemini_api_key_here
   

4. Build the project

   npm run build
   

Running the Application

1. Start the client

   node build/client.js
   

2. Choose an option from the menu:

   • Query (AI-powered): Ask questions in natural language (e.g., "Create a user named John with email john@example.com")
   • Call a Tool: Manually execute tools like creating users
   • Read a Resource: View all users or specific user details
   • Use a Prompt: Generate fake user data based on templates
   • Exit: Close the application

📁 Project Structure

├── src/
│   ├── client.ts          # MCP client implementation
│   ├── server.ts          # MCP server implementation
│   └── data/
│       └── users.json     # User database (auto-created)
├── build/                 # Compiled JavaScript files
├── .env                   # Environment variables
├── package.json
└── README.md

🔧 Available Features

Resources

• Get All Users (users://all): Retrieve all users from the database
• Get User Details (users://{userId}/profile): Get specific user information

Tools

• create-user: Add a new user with name, email, address, and phone
• create-random-user: Generate and save a fake user using AI

Prompts

• generate-fake-user: Template for creating fake user data based on a name

💡 Example Usage

Using AI Query

What would you like to do? > Query (AI-powered)
Enter your query: > Create a new user named Alice with email alice@example.com

Reading All Users

What would you like to do? > Read a Resource
Select a resource: > Users (users://all)

Creating a Random User

What would you like to do? > Call a Tool
Select a tool: > Create Random User

🛠️ Development

Build the project

npm run build

Watch mode (auto-rebuild on changes)

npm run build -- --watch

📦 Dependencies

• @ai-sdk/google: Google Gemini AI integration
• @modelcontextprotocol/sdk: MCP protocol implementation
• ai: Vercel AI SDK for unified AI interactions
• @inquirer/prompts: Interactive CLI prompts
• zod: Schema validation
• dotenv: Environment variable management

🤝 Contributing

Contributions are welcome! Feel free to:

• Report bugs
• Suggest new features
• Submit pull requests

📄 License

[Add your license here]

🔗 Resources

• MCP Documentation
• Google Gemini API
• Vercel AI SDK

⚠️ Notes

• The user database is stored in src/data/users.json and persists between sessions
• User IDs are auto-incremented
• The server uses stdio transport for communication
• All AI operations use Google's Gemini 2.0 Flash model