mohamed-elkholy95/mcp-project
Educational from-scratch implementation of Anthropic's Model Context Protocol (MCP) with tools, resources, prompts, and interactive dashboard
Platform-specific configuration:
{
"mcpServers": {
"mcp-project": {
"command": "npx",
"args": [
"-y",
"mcp-project"
]
}
}
}Add the config above to .claude/settings.json under the mcpServers key.
<div align="center">
An educational, from-scratch implementation of Anthropic's Model Context Protocol
[](https://python.org) [](#) [](https://fastapi.tiangolo.com) [](https://streamlit.io)
</div>
Model Context Protocol (MCP) is an open-source standard by Anthropic for connecting AI applications to external systems. Think of it as a USB-C port for AI — a universal connector that lets any AI model (Claude, ChatGPT, Copilot) talk to any data source, tool, or workflow.
┌──────────────────┐ ┌──────────────────┐
│ Claude Desktop │ JSON │ MCP Server │
│ ┌────────────┐ │ RPC │ ┌────────────┐ │
│ │ MCP Client │←─┼────────→│ │ Tools │ │
│ └────────────┘ │ 2.0 │ │ Resources│ │
│ │ │ │ Prompts │ │
└──────────────────┘ └──────────────────┘| Primitive | Purpose | Example | |-----------|---------|---------| | Tools | Actions the AI can take | Calculator, file operations, API calls | | Resources | Data the AI can read | Config files, documents, database records | | Prompts | Templates for interactions | Code review template, data analysis guide | | Transport | Communication channel | Stdio (local) or HTTP (remote) |
This is a fully educational, heavily commented implementation that shows how MCP works under the hood — not using the SDK, but building the concepts from scratch:
Loading reviews...