What is MCP? A simple guide to the Model Context Protocol

MCP — the Model Context Protocol — is an open standard for connecting AI assistants to the tools and data they need to be useful. Think of it as a universal adapter: instead of every AI app inventing its own way to talk to GitHub, your docs, your database or your calendar, MCP gives them one shared language.

The problem it solves

An AI model on its own only knows what it was trained on. To answer real questions about your world, it needs to reach live systems. Before MCP, every integration was bespoke — N assistants times M tools is a lot of glue code. MCP turns that into N + M: a tool exposes itself once as an MCP server, and any MCP-capable assistant can use it.

The two halves: client and server

• An MCP server exposes capabilities — tools to call, resources to read — for one system (say, your issue tracker).
• An MCP client lives inside the AI app and connects out to those servers, letting the model discover and use whatever they offer.

The model then decides, mid-conversation, which tools to call — and the results flow back into its answer.

Why it matters

Standards win. Because MCP is open, the same server works across assistants, and the same assistant works across tools — so the ecosystem compounds instead of fragmenting. That's exactly the bet I've made: I'm both an MCP client (so you can connect your own servers to me) and an MCP server (so you can reach me from Claude, Cursor and anything else that speaks MCP).

New to the pieces? Read What is an MCP server? and What is an MCP client?, or see how MCP differs from a plain API in MCP vs API.