MCP Servers Learning Project
@sagarnildass
MCP Crash Course: Complete Model Context Protocol in a Day course from Udemy
This repository contains examples and implementations of Model Context Protocol (MCP) servers and clients. MCP is a protocol that allows LLMs like Claude to interact with external tools and data sources in a standardized way.
Project Structure
The repository is organized into several modules, each focusing on different aspects of MCP:
1. Quickstart Resources
The 1_quickstart-resources directory contains sample implementations of MCP clients and servers to help you get started:
- mcp-client-python: A Python client implementation for connecting to MCP servers
- mcp-client-typescript: A TypeScript client implementation for connecting to MCP servers
- weather-server-python: A sample Python MCP server that provides weather information
- weather-server-typescript: A sample TypeScript MCP server that provides weather information
These examples demonstrate basic MCP concepts and provide starting points for your own implementations.
2. MCP Documentation Server
The 2_mcpdoc directory contains a specialized MCP server for serving documentation:
- Implements an MCP server that can retrieve and serve documentation from various sources
- Includes configuration examples and a robust project structure
- Demonstrates advanced MCP concepts like error handling and resource management
3. Terminal Server
The 3_building_securing_and_containerizing_mcp_server directory contains a custom MCP server implementation that:
- Exposes a terminal command execution tool allowing access to system commands
- Provides a resource for accessing MCP documentation
- Demonstrates proper error handling and input validation
- Shows how to implement both tools and resources in a single server
Getting Started
Prerequisites
- Python 3.10 or higher
- MCP SDK 1.2.0 or higher
- For TypeScript examples: Node.js and npm
Installation
# For Python dependencies
uv add "mcp[cli]" httpx
# Or with pip
pip install "mcp[cli]" httpx
Running MCP Servers
You can run any of the MCP servers using:
# With Python directly
python <path_to_server.py>
# Or with the MCP CLI
mcp dev <path_to_server.py>
For example, to run the terminal server:
python 3_building_securing_and_containerizing_mcp_server/server.py
# Or
mcp dev 3_building_securing_and_containerizing_mcp_server/server.py
Connecting to Claude Desktop
To connect an MCP server to Claude Desktop:
mcp install <path_to_server.py>
Key Concepts
- Tools: Functions that can be called by Claude to execute code or produce side effects
- Resources: Data exposed to Claude that can be read and referenced
- Prompts: Templates for structuring interactions with Claude
Contributing
Feel free to extend and modify these examples for your own use cases. If you develop a useful MCP server, consider sharing it with the community!
