What is MCP (Model Context Protocol)?
MCP is an open-source standard for connecting AI applications to external systems: data sources such as files and databases, tools such as search engines and calculators, and workflows such as specialized prompts.
The official introduction compares it to a USB-C port for AI applications. USB-C gives devices one standard way to connect. MCP gives AI applications one standard way to reach external systems.
Architecture
- Host: the AI application, for example Claude Code or Visual Studio Code. It manages the clients.
- Client: a component inside the host that keeps one connection to one server.
- Server: a program that provides context to clients. It can run locally or remotely.
Primitives a server can expose
- Tools: executable functions the AI application can invoke, such as file operations, API calls and database queries.
- Resources: data that gives context, such as file contents or database records.
- Prompts: reusable templates, such as system prompts and few-shot examples.
Clients discover what a server offers through list methods, for example tools/list, and run a tool
with tools/call.
Transports
MCP defines two: stdio for local processes, and Streamable HTTP for remote servers. Messages are JSON-RPC 2.0.
Version note
The architecture page read on 2026-10-20 describes protocol version 2026-07-28, where requests carry the
protocol version and capabilities and discovery uses server/discover. Earlier versions work differently (for
example, with an initialize handshake). Check the version your SDK targets.
MCP and A2A
MCP is about an agent using capabilities. It does not define how agents hand work to each other. That is A2A's job. See the comparison and the demo, which includes one MCP tool call next to the A2A traffic.
FAQ
- What is MCP?
- MCP (Model Context Protocol) is an open-source standard for connecting AI applications to external systems such as data sources, tools and workflows.
- What are the main MCP primitives?
- Servers expose three core primitives: tools (functions the model can invoke), resources (context data) and prompts (reusable templates).
- Why is MCP compared to USB-C?
- The official introduction says MCP is like a USB-C port for AI applications: one standard way to connect to many external systems instead of a custom connector for each.
- Does MCP connect agents to other agents?
- That is not its purpose. MCP connects an AI application to tools and context. Agent to agent collaboration is the job of A2A.