MCP (Model Context Protocol)

What is MCP?

Model Context Protocol (MCP) is an open standard created by Anthropic that provides a universal way for AI models to interact with external systems. Think of it as USB for AI — a single protocol that connects any AI model to any tool, database, or API.

How Does MCP Work?

MCP uses a client-server architecture:

1. MCP Server: Wraps an external tool or service (database, API, file system) and exposes its capabilities
2. MCP Client: The AI application (like Claude Code) that discovers and uses available servers
3. Tool Discovery: The client automatically learns what tools are available and how to use them
4. Execution: The AI calls tools through the MCP protocol, receives results, and uses them in its responses

Why MCP Matters

Before MCP	After MCP
Each AI tool needs custom integration code	Universal protocol — write once, use everywhere
Limited to what the AI knows from training	Access to live data, APIs, and real-time systems
AI can only generate text	AI can take real-world actions
Fragmented ecosystem	Interoperable standard

MCP in Practice

Claude Code uses MCP extensively — connecting to Supabase databases, Google Search Console, GitHub, Playwright browsers, and dozens of other tools. Each MCP server gives Claude Code new capabilities without any custom coding.

Key Takeaway: MCP is the infrastructure layer that makes AI agents practical — without it, AI models are limited to generating text. With it, they can interact with the real world.