Model Context Protocol

The Model Context Protocol, or MCP, is an open standard for connecting AI applications to outside tools and data. Anthropic introduced it, and it is now supported by many other AI apps, coding tools and agent frameworks. The idea is that a tool is wrapped once as an MCP server and then works with any client that speaks the protocol.

An MCP server can expose tools the model can call, resources it can read and prompt templates. Official SDKs exist for several languages, and there are servers for file systems, databases, code hosts, chat apps and many business tools. Servers can run on your own machine or as a remote service.

For builders, MCP cuts the work of writing a separate integration for each agent. For operators, it is also a security question: a server runs with whatever access you give it. Review a server's code or publisher before installing it, grant the narrowest credentials that work, and keep human approval on actions that change data.