Glossary
What is Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open standard, introduced by Anthropic in November 2024, that defines how AI applications connect to external tools and data sources. It gives models a uniform way to call functions and read resources, replacing the bespoke integration each vendor previously wrote for every tool.
Last updated:
Before MCP, connecting a model to a CRM, a database or a file store meant writing a one-off adapter for each pair of model and tool — an integration matrix that grew multiplicatively. MCP inverts that: a tool exposes an MCP server once, and any MCP-capable client can use it.
The protocol defines three primitives. Tools are functions the model can invoke. Resources are data the model can read. Prompts are reusable templates a server can offer. Transport runs over stdio for local processes or HTTP for remote servers.
MCP is frequently confused with the A2A protocol, and the distinction matters for anyone designing agent systems: MCP connects a model to its own capabilities, while A2A connects independent agents to each other. An agent typically speaks both — MCP inward to its tools, A2A outward to other organisations' agents. AgentDoor sits at the A2A layer: it is concerned with how a buyer's agent and a vendor's agent hold a structured conversation, not with how either one reaches its internal systems.
Frequently asked questions
- Is MCP a competitor to the A2A protocol?
- No — they operate at different layers and are commonly used together. MCP standardises a model's access to tools and data; A2A standardises communication between separate agents.
Skip the discovery calls.
Describe what you need and let AgentDoor's agents interview the vendors for you — you get a decision-ready shortlist, not six meetings.
We'll reach out at launch. No spam, ever.