Sagnik-23/MCP_Server

MCP Server and Client

This project is a Model Context Protocol (MCP) server and client implementation in TypeScript. It allows you to run an MCP server, interact with it using a client, and manage resources and tools such as user creation and querying.

Features

• MCP server with resource and tool definitions
• User management (create, list, etc.)
• Interactive client with prompts for tools, resources, and queries
• Integration with Google Generative AI (Gemini)
• TypeScript-based codebase

Getting Started

Prerequisites

• Node.js (v18+ recommended)
• npm

Installation

1. Clone the repository:

   git clone <repo-url>
   cd MCP_Server
   

2. Install dependencies:

   npm install
   

Build

To compile TypeScript to JavaScript:

npm run server:build

Run the Server

To start the MCP server:

npm run server:dev

Run the Client

To start the interactive client:

npm run server:dev

Usage

• The client will prompt you to select actions: Query, Tools, Resources, Prompts, or Exit.
• For tools (e.g., user creation), enter the required fields as prompted.
• For resources, select and provide any required parameters.
• For prompts, interact with AI-powered text generation.

Project Structure

MCP_Server/
├── src/
│   ├── server.ts      # MCP server implementation
│   ├── client.ts      # Interactive client
│   └── data/
│       └── users.json # User data storage
├── build/             # Compiled JS output
├── package.json       # Project configuration
├── tsconfig.json      # TypeScript configuration
└── README.md          # Project documentation

Environment Variables

• To use Google Generative AI, set your API key in a .env file:

  GEMINI_API_KEY=your-google-api-key
  

License

ISC

Author

Sagnik-23