Learn MCP

Evidence-backed fundamentals, protocol architecture, tools, resources, and implementation guides for the Model Context Protocol.

MCP Fundamentals

Model Context Protocol (MCP) explained: what it is, why it exists, how it works, and the host/client/server architecture.

What an MCP server is, what it exposes, how local and remote servers differ, and how to choose and run one.

What an MCP client is, the client lifecycle, tool and resource calls, and which applications function as clients.

The MCP three-layer architecture: host, client, and server, plus the request lifecycle, capability negotiation, and transports.

What MCP tools are: the tools/list and tools/call lifecycle, tool schemas, arguments, execution, and security.

What MCP resources are, resource URIs, the resources/list request, and how resources differ from tools.

What MCP prompts are, the prompts/list and prompts/get lifecycle, prompt arguments, and how prompts differ from tools.

The MCP protocol: JSON-RPC message layer, initialization, capability negotiation, primitives, and transports.

Real MCP use cases across development workflows, databases, browser automation, research, and enterprise systems.

A step-by-step explanation of how an MCP request flows from host to client to server and back, with common failure points.

MCP uses JSON-RPC 2.0 as its message layer: request/response/notification structure, batch requests, and error codes.

The stdio transport runs the server as a child process and exchanges JSON-RPC over standard input and output.

The Streamable HTTP transport runs MCP over HTTP with server-sent events for server-to-client messages and POST for client-to-server.