mcp-server-trail-project
If you are the rightful owner of mcp-server-trail-project and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.
This project demonstrates integration between a client application and an MCP (Model Context Protocol) server, allowing for AI-powered interactions with tool execution capabilities.
Model Context Protocol (MCP) Server Project
This project demonstrates integration between a client application and an MCP (Model Context Protocol) server, allowing for AI-powered interactions with tool execution capabilities.
Project Overview
This application consists of two main components:
- A client that connects to Google's Gemini AI model and an MCP server
- An MCP server that registers and provides tools for the AI model to use
The system allows users to interact with the Gemini AI model through a command-line interface. The AI can respond to user queries and execute specialized tools hosted on the MCP server, such as posting tweets or performing calculations.
Architecture
├── client/ # Client application
│ ├── .env # Environment variables for client
│ ├── index.js # Client implementation
│ └── package.json # Client dependencies
└── server/ # MCP server
├── .env # Environment variables for server
├── index.js # Server implementation
├── mcp.tool.js # Tool implementations
└── package.json # Server dependencies
Features
- AI chat interface using Google's Gemini model
- Tool execution through MCP protocol
- Available tools:
addTwoNumbers: Performs addition of two numberscreatePost: Creates a post on Twitter/X
Setup and Installation
Prerequisites
- Node.js (v14 or higher)
- npm or yarn
- Twitter/X API credentials
Server Setup
- Navigate to the server directory:
cd server - Install dependencies:
npm install - Configure the
.envfile with your Twitter API credentials:TWITTER_API_KEY=your_api_key TWITTER_API_KEY_SECRET=your_api_secret TWITTER_ACCESS_TOKEN=your_access_token TWITTER_ACCESS_TOKEN_SECRET=your_access_token_secret - Start the server:
node index.js
Client Setup
- Navigate to the client directory:
cd client - Install dependencies:
npm install - Configure the
.envfile with your Gemini API key:GEMINI_API_KEY=your_gemini_api_key - Start the client:
node index.js
Usage
- Start the server first, then the client
- When the client connects, you'll see a prompt for input
- Type your question or request
- The AI will respond directly or use one of the tools if needed
Example interactions:
- "What's 25 plus 17?" (Uses the addTwoNumbers tool)
- "Post a tweet that says 'Hello from my MCP project!'" (Uses the createPost tool)
How It Works
- The client connects to the MCP server via SSE (Server-Sent Events)
- The server registers available tools with input schemas using Zod validation
- User queries are sent to Google's Gemini AI model
- If the AI determines a tool should be used, it makes a function call
- The function call is routed through the MCP client to the MCP server
- The server executes the requested tool and returns results
- Results are presented to the user and added to chat history
Technologies Used
- @modelcontextprotocol/sdk - For MCP implementation
- @google/genai - For Gemini AI integration
- Express - Web server framework
- twitter-api-v2 - Twitter API client
- zod - Schema validation
