Vocabulary
MCP
Model Context Protocol: a shared way for AI apps to plug into tools and data, so every pair does not need its own adapter.
Also called: Model Context Protocol
MCP is an open protocol for connecting AI applications to external tools and data sources. Think USB-C for model tooling: one shape of plug, many devices.
Before MCP, wiring a chat app to your notes, calendar, or homelab often meant a custom connector for each pair. MCP aims to replace that drawer of adapters with a common handshake: the app speaks MCP, the tool exposes an MCP server, and they can talk without a one-off script for every combination.
It does not replace APIs. Under the hood, tools still do real work through ordinary software interfaces. MCP is the agreed etiquette layer so models and apps can discover and call those capabilities more uniformly.
When someone says “we need MCP” in a meeting, the useful follow-up is usually: which tools should the model reach, and who runs the servers that expose them?
Cups that use this term