MCP Architecture: How the Model Context Protocol Works

Youve heard about MCP. You know it connects Claude to external tools and data sources. But how does it actually work under the hood?

This is Part 2 of our Understanding MCP series. In Part 1, we covered what MCP is and why it matters. Now were diving into MCP architecture—the technical foundation that makes it all possible.

The Client-Server Model

MCP uses a straightforward client-server architecture. MCP Servers are services that expose capabilities—data, tools, or prompt templates. MCP Hosts (clients) are applications that connect to these servers.

Claude Desktop is an MCP host. It can connect to multiple MCP servers simultaneously, giving Claude access to all their combined capabilities.

Transport Layers

stdio Transport

The stdio transport runs MCP servers as local processes communicating via standard input/output streams. Perfect for development and single-user scenarios.

HTTP/SSE Transport

For production, MCP uses HTTP with Server-Sent Events. Servers can run anywhere, and multiple clients can connect to the same server.

The JSON-RPC Protocol

MCP builds on JSON-RPC 2.0. Every message follows this structure:

{
  jsonrpc: 2.0,
  id: 1,
  method: tools/call,
  params: { name: read_file, arguments: { path: /file.txt } }
}

Core Primitives

Resources

Resources represent data servers can provide—files, database records, API responses. Theyre identified by URIs and are read-only.

Tools

Tools are functions the model can call to perform actions. They can have side effects—creating files, sending messages, updating databases.

Prompts

Prompts are reusable templates with optional arguments, useful for standardizing common workflows.

Capability Negotiation

Before operations begin, client and server agree on supported capabilities via an initialization handshake. This ensures compatibility.

Message Flow

User asks Claude a question requiring external data
Claude requests resources/tools from the MCP server
Server executes the operation and returns results
Claude incorporates data into its response

Coming Next

In [Part 3](/blog/build-mcp-server-tutorial), well build a working MCP server from scratch with complete code examples.

How This Article Was Made

This article is a live example of the AI-enabled content workflow we build for clients.

Stage	Who	What
Research	Claude Opus 4.5	Analyzed current industry data, studies, and expert sources
Curation	Tom Hundley	Directed focus, validated relevance, ensured strategic alignment
Drafting	Claude Opus 4.5	Synthesized research into structured narrative
Fact-Check	Human + AI	All statistics linked to original sources below
Editorial	Tom Hundley	Final review for accuracy, tone, and value

The result: Research-backed content in a fraction of the time, with full transparency and human accountability.

Why We Work This Way

Were an AI enablement company. It would be strange if we didnt use AI to create content. But more importantly, we believe the future of professional content isnt AI vs. Human—its AI amplifying human expertise.

Every article we publish demonstrates the same workflow we help clients implement: AI handles the heavy lifting of research and drafting, humans provide direction, judgment, and accountability.

Want to build this capability for your team? Lets talk about AI enablement →