What is MCP (Model Context Protocol)?

An open standard that lets AI assistants connect to external tools, data sources and apps through a common interface.

MCP (Model Context Protocol) is an open standard, introduced by Anthropic in late 2024, that defines a common way for AI assistants to connect to external tools and data. Before MCP, every AI app needed custom integration code for each service it wanted to touch. With MCP, a service exposes an "MCP server" that describes its tools and resources in a standard format, and any MCP-compatible assistant (an "MCP client") can use it. One integration works across many AI apps. Servers exist for things like GitHub, Google Drive, Slack, databases, and web browsers, and the standard has been adopted beyond Anthropic, including by OpenAI and Google. Practical nuance: MCP handles the plumbing (discovery, connection, message format); the model still decides when and how to use each tool, so quality depends on both the server and the model.

Example

You add a GitHub MCP server to your AI coding assistant. Without any custom code, the assistant can now list your open pull requests, read issue threads, and comment on them, using the same server another MCP-compatible app could use.

Why it matters

MCP support is becoming a checklist item for AI tools. A tool that speaks MCP can plug into a growing ecosystem of ready-made connectors instead of a short list of vendor-built integrations. Browse the AI tools directory or the model leaderboard to put it into practice.