Glossary
What is the Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open standard, introduced by Anthropic in November 2024, for connecting AI applications to external tools and data. An MCP server exposes tools, resources and prompts; an MCP client, such as Claude, Cursor or ChatGPT, connects and lets the model use them.
How it works
| Part | Role |
|---|---|
| Host / client | The AI app (Claude, Cursor, ChatGPT and others) |
| Server | A program exposing capabilities |
| Tools | Actions the model can call |
| Resources | Data the model can read |
| Prompts | Reusable templates |
Servers run locally (over standard input/output) or remotely (over HTTP), and remote servers can use OAuth for authorisation.
Why it caught on
Before MCP, every AI app needed custom integrations for every tool. A shared protocol means a tool built once works across many clients. Major AI companies and developer tools adopted it within a year of release.
Where to find servers
MCP registries list servers; an official registry lets publishers register servers under verified namespaces. Our list of MCP registries covers the main ones.
Example
A developer adds a database MCP server to her coding agent. The agent can now query tables directly when asked about the data, instead of her copying results into chat.
FAQ
Who created MCP?
Anthropic, in November 2024, as an open standard.
Is MCP only for Claude?
No. Many AI apps and tools support it.
Is MCP secure?
It depends on the servers you install; use trusted ones and review permissions.