MCP Demo - TypeScript Implementation

This is a TypeScript implementation of the MCP: Build Rich-Context AI Apps with Anthropic course from DeepLearning.AI.

Overview

This project demonstrates the Model Context Protocol (MCP) implementation with streamable HTTP capabilities. MCP is an open protocol that standardizes how LLM applications can access context through tools and data resources using a client-server architecture.

⚠️ This project is for educational and demo purposes only.

Features

• MCP client-server architecture implementation
• Streamable HTTP communication
• arXiv paper search functionality
• Paper information extraction
• Tool selection and argument extraction
• Prompt template management

Prerequisites

• Node.js (v16 or higher)
• Yarn package manager
• Anthropic API key

Setup

1. Clone the repository

   git clone <repository-url>
   cd mcp-demo
   

2. Install dependencies

   yarn install
   

3. Environment Configuration

   Create a .env file in the root directory:

   ANTHROPIC_API_KEY=<your_anthropic_api_key_here>
   

   Important: Replace <your_anthropic_api_key_here> with your actual Anthropic API key.

4. Build the project

   yarn build
   

Project Structure

mcp-demo/
├── src/
│   ├── client.ts      # MCP client implementation
│   ├── server.ts      # MCP server implementation
│   └── index.ts       # Core functionality and utilities
├── package.json       # Dependencies and scripts
├── tsconfig.json      # TypeScript configuration
├── yarn.lock          # Locked dependencies
└── README.md          # This file

Usage

Starting the MCP Server

yarn start:server

Starting the MCP Client

yarn start:client

Running Both (Development)

yarn dev

Available Tools

The MCP server provides the following tools:

1. search_papers - Search for papers on arXiv

   • Arguments:
     • topic (string): The topic to search for
     • max_results (number, optional): Maximum number of results (default: 5)

2. extract_info - Extract information from a specific paper

   • Arguments:
     • paper_id (string): The ID of the paper to look for

API Reference

search_papers(topic: string, max_results?: number)

Searches for papers on arXiv based on a topic and returns their information.

extract_info(paper_id: string)

Searches for information about a specific paper by ID from arXiv.

getToolSelectionPrompt(toolList: string, userQuery: string)

Generates a detailed prompt for tool selection and argument extraction.

Course Reference

This implementation is based on the MCP: Build Rich-Context AI Apps with Anthropic course by DeepLearning.AI in partnership with Anthropic. The course covers:

• Core concepts of MCP
• Client-server architecture
• Building MCP-compatible applications
• Connecting to third-party servers
• Deploying MCP servers remotely

For the complete course content, visit: https://learn.deeplearning.ai/courses/mcp-build-rich-context-ai-apps-with-anthropic

Contributing

This is a demo project for educational purposes. Feel free to experiment and modify the code to learn more about MCP implementation.

License

This project is for educational purposes only. Please refer to the original course materials for licensing information.

Support

For questions about the MCP protocol or the original course, please refer to:

• MCP Documentation
• DeepLearning.AI Course
• Anthropic Documentation