MCP Practice Project

A Model Context Protocol (MCP) server and client implementation for learning and practicing MCP development. This project demonstrates how to build an MCP server that exposes resources, tools, and prompts, along with an interactive CLI client to interact with them.

Features

MCP Server (src/server.ts):

• Resources: Query user data via users://all and users://{userId}/profile
• Tools:
  • create-user - Create a user with explicit fields (name, email, address, phone)
  • create-random-user - Generate a random user using LLM sampling
• Prompts: generate-fake-user - Generate fake user data based on a name

MCP Client (src/client.ts):

• Interactive CLI for testing MCP server capabilities
• Query tools, resources, and prompts
• Natural language query interface using Google Gemini AI

Tech Stack

• Runtime: Node.js v24.10.0
• Language: TypeScript (strict mode, ES modules)
• Framework: Model Context Protocol SDK (@modelcontextprotocol/sdk)
• AI: Google Gemini API (@ai-sdk/google)
• Validation: Zod
• Storage: JSON file (src/data/users.json)

Quick Start

# Install dependencies
npm install

# Build the server
npm run server:build

# Run the server (for MCP Inspector)
npm run server:inspect

# Run the interactive client
npm run client:dev

Project Structure

src/
  ├── server.ts      # MCP server implementation
  ├── client.ts      # MCP client CLI
  └── data/
      └── users.json # JSON-based user database

Development

This is a practice/learning project for understanding MCP development patterns. The server uses StdioServerTransport for communication, and data is stored in a simple JSON file (not production-ready).