Glossary
Model Context Protocol
Model Context Protocol (MCP) is an open standard for connecting AI applications to outside tools and data, such as files, databases, task trackers or other software. Instead of building a separate integration for every pair, an app that speaks MCP can use any MCP server that exposes those tools.
How Model Context Protocol works
MCP was introduced by Anthropic in November 2024 as an open protocol. An AI app acts as a client, and each connected system runs an MCP server that describes what it offers, such as tools the model can call and data it can read. The model then uses them through one common interface.
An example of MCP
A team connects its AI assistant to the CRM and the shared drive through MCP servers. Asked "what did we promise this client", the assistant reads the client's record and the signed proposal and answers from live data, without anyone copying them into the chat.
Is MCP secure?
MCP is as secure as the servers you connect and the access you give them. A server can read data and take actions, so the usual rules apply: use trusted servers, give each the least access it needs, keep a person approving risky actions and watch for instructions hidden in content.