mcp_starter

The MCP Starter Project is designed to facilitate the development of AI applications that can interact with external tools and APIs using the Model Context Protocol (MCP). It consists of two main components: the MCP Server, a Python service that defines and exposes tools/functions callable by AI models, and the MCP Client, a TypeScript/JavaScript client that connects to the MCP server to manage interactions between AI models and tools. The project structure includes a Python MCP server implementation and a TypeScript MCP client implementation, with integration for Hugging Face and Google Search functionalities. The server and client are set up with specific dependencies and environment variables, and they communicate through defined endpoints for health checks and chat interactions.

Features

• {'name': 'Documentation Search Tool', 'description': 'Search documentation for popular AI libraries such as LangChain, LlamaIndex, and OpenAI.'}
• {'name': 'Conversation Management', 'description': 'Maintains chat history per session to ensure context is preserved.'}
• {'name': 'Tool Integration', 'description': 'Seamlessly integrates AI model responses with tool calls for enhanced functionality.'}
• {'name': 'Error Handling', 'description': 'Robust error handling for API calls and tool execution to ensure reliability.'}

Usage with Different Platforms

mcp_server_setup

bash
cd mcp-server
python -m venv .venv
# On Windows
.venv\Scripts\activate
pip install -e .

mcp_client_setup

bash
cd mcp-clients
npm install
npm run build

start_mcp_server

bash
cd mcp-server
python main.py

start_mcp_client

bash
cd mcp-clients
node build/index.js ../mcp-server/main.py