MCP (Model Context Protocol)
An open standard that lets AI models securely connect to external tools, databases, and services.
The Model Context Protocol (MCP) is an open standard, published by Anthropic in 2024, that defines how AI models connect to external data sources and tools. It provides a consistent interface so that an AI assistant can query a database, call an API, read a file system, or interact with any service — without each integration needing to be custom-built.
Before MCP, connecting an AI to your tools required writing bespoke "tool use" code for every integration. MCP standardises this: tools are MCP servers, AI models are MCP clients, and the protocol defines how they discover capabilities, pass data, and maintain security.
For vibe coders, MCP is increasingly relevant as agentic workflows mature. Understanding MCP means being able to build your own custom tool integrations that AI agents can use — extending what the AI can access beyond the defaults.