Learn MCP

What is MCP?

MCP — the Model Context Protocol — is an open standard published by Anthropic in late 2024. It defines a universal way for AI assistants to connect to external tools, databases, APIs, and services. Think of it as a USB standard, but for AI: any MCP-compatible tool works with any MCP-compatible AI client, without custom glue code.

How MCP Works

MCP has three moving parts: a client (your AI app), a server (the integration), and the protocol connecting them. Understanding how they interact helps you choose servers wisely and troubleshoot problems when they arise.

Why MCP Exists

Before MCP, every AI product needed a custom integration for every tool. M AI products × N tools = M×N bespoke connections to build and maintain. MCP collapses that to M + N.

MCP vs. Plugins & Function Calling

Three approaches exist for giving AI access to external tools: OpenAI function calling, ChatGPT plugins (now largely deprecated), and MCP. Here's how they differ.

MCP Clients Guide

An MCP client is the AI application that connects to your MCP servers. Each client has its own config file location and feature set. Here's what you need to know about each.

Anatomy of an MCP Server

You don't need to build an MCP server to use one, but understanding the structure helps you evaluate servers, read their source code, and eventually build your own.

MCP Tools vs. Resources

MCP servers can expose two kinds of capabilities: tools and resources. Tools are actions — the AI calls them to do something. Resources are data — the AI reads them to know something. Most servers today focus on tools, but resources are powerful for read-heavy use cases.

Getting Started with MCP

This guide walks you through installing the filesystem MCP server — one of the most useful servers for developers — in Claude Desktop on macOS. The same pattern applies to any server and any client.