What is MCP (Model Context Protocol)?
MCP is a common plug between AI applications and the things they need to use.
Read on 14 November 2026. This field changes fast: check the sources.
In the project's words
The MCP documentation says that MCP (Model Context Protocol) is an open-source standard for connecting AI applications to external systems. With MCP, AI applications such as Claude or ChatGPT can connect to data sources (for example local files and databases), tools (for example search engines and calculators) and workflows (for example specialized prompts), so that they can reach key information and perform tasks. Its own comparison: MCP is like a USB-C port for AI applications. Just as USB-C gives a standard way to connect electronic devices, MCP gives a standard way to connect AI applications to external systems.
Why a standard helps
Without a standard, every AI application needs its own code for every service. With one, a service can offer an MCP server once, and any application that speaks MCP can use it. The documentation lists examples of what this enables, such as agents using a calendar and notes, or a coding tool building a web app from a design file.
What MCP is not
It is not a model and it does not make a model smarter. It does not decide what is safe: the application and the server author must do that. And it is not the only way: you can define tools directly in your own program, as the page on AI agents shows.
Where to go next
The next page describes the parts (host, client, server), the layers and the three kinds of things a server can offer. The page after that builds a small server and shows the real messages.
Who runs MCP?
It is an open-source project with public documentation and a specification. The specification versions are dated; the site's addresses showed 2026-07-28 when we read it.
Is it only for Claude?
The documentation names several AI applications as examples, and describes MCP as open.
Sources
- Model Context Protocol documentation: What is MCP? (modelcontextprotocol.io/docs/getting-started/intro)
Read on 14 November 2026.
This is an independent guide. It is not connected with Anthropic or the MCP project. This field changes quickly: statements were read on 14 November 2026 from the pages named under Sources (the MCP site shows specification version 2026-07-28 in its page addresses). The demo server and the messages are real: we ran them with the official Python SDK (mcp 2.3.0). No AI model was used for the demos.