Model Context Protocol (MCP)

The Model Context Protocol (MCP) is an open protocol that standardizes how AI applications connect to external tools, data sources and prompt templates.

Origin and governance. Anthropic introduced MCP in late 2024. In December 2025 it donated the protocol to the Agentic AI Foundation, a directed fund under the Linux Foundation. The specification is versioned by date, and at the time of writing the current revision is 2026-07-28.

How it works. MCP uses a host, client and server model. The host is the AI application. It runs one MCP client per connection, and each client talks to one MCP server. Servers offer three kinds of features:

• Tools: functions the model can call, such as a database query or a ticket lookup.
• Resources: context and data for the user or the model.
• Prompts: templated messages and workflows.

Messages are JSON-RPC 2.0. The two standard transports are stdio, for a local server started as a subprocess, and Streamable HTTP, for remote servers. The 2026-07-28 revision made MCP stateless. It removed the initialization handshake, so every request now carries its own protocol version and client capabilities.

Relation to A2A. MCP deepens what a single agent can do. A2A connects agents that remain separate systems. The A2A specification gives the usual combination: a client agent sends a task to a remote agent over A2A, and the remote agent uses MCP internally to reach the tools and data it needs to finish the task. The A2A project docs also caution against wrapping a whole agent as an MCP tool. Tools usually have fixed inputs and outputs, while agents keep state across multi-turn conversations.

Neighbouring terms. JSON-RPC 2.0 is the message format for MCP and for A2A’s JSON-RPC binding. The x402 payment standard defines transports for both MCP and A2A.