What is the Model Context Protocol (MCP)?

The Model Context Protocol (MCP) is an open protocol that lets AI applications such as Claude Desktop, Claude Code and other clients call tools and read data provided by separate programs called MCP servers. A server might search your files, query a database or read your research library; the client decides when to call it and sends what it returns to the model.

An MCP server runs locally or remotely and describes its tools with names, inputs and descriptions the model can read.

What a tool returns becomes part of the conversation, so it goes wherever the conversation goes, for Claude Desktop to Anthropic.

Read next

The whole glossary →

Help PaperLab reach the next researcher

PaperLab is free, open source and built by one person. A star helps other researchers find it and puts each new release in your GitHub feed. The PaperLab page on LinkedIn posts a weekly breakdown of how one feature works.