Introduction to Model Context Protocol (MCP)

Conceptual overview, no coding here, just the terminology and mental model. Assumes general familiarity with LLMs; no prior MCP knowledge required.

The problem MCP solves

A model's training data alone isn't enough for real-world AI applications. Often the AI needs access to information outside itself, a user's calendar, a database, files in a file system, emails in an inbox, or any other external source. The open question is: how do you connect that information up to the AI in a standardized way? That's what MCP answers.

Breaking down the name

• Standard: a shared agreement on how something should work, so things stay compatible. Not a technology in itself, but an agreed-upon way of doing something.
• Protocol: a set of rules for communication: the format, order, and meaning of messages exchanged between an AI application and external information/tools.
• Open (open source), publicly available for anyone to use or build on. MCP was created by Anthropic, but the protocol itself is open, anyone can implement and use it.

Definition

MCP is an open-source standard protocol for connecting AI applications to external tools and data. The rest of the course/topic goes into the details of that protocol, and how tools and data are actually made available to an AI through it.