MCP Servers

What Is MCP?

The Model Context Protocol (MCP) is an open standard created by Anthropic that lets AI models connect to external data sources and tools through a universal interface. A protocol is simply an agreed-upon set of rules for how two systems communicate — like how HTTP defines how your browser talks to websites. MCP defines how AI models talk to tools.

Instead of building custom integrations for every tool, MCP provides a single protocol that any server can implement. Think of it like USB for AI — one standard connector that works with everything.

Why this matters: Without MCP, connecting Claude to your files, databases, or APIs would require custom tool definitions for each one. With MCP, you write a server once using the standard protocol, and any MCP-compatible client (Claude Code, Claude Desktop, Cursor, and more) can use it automatically.

Architecture Overview

The MCP architecture has three layers:

The flow is: Client discovers server tools at startup → AI model decides to call a tool → Client sends the call to the server → Server executes against data sources → Result flows back to the AI model.

Building a Simple MCP Server

Here is a complete, working MCP server in TypeScript. This server exposes a single tool — a note-taking system. You can run this with Claude Code or Claude Desktop:

// npm install @modelcontextprotocol/sdk
import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import { z } from "zod";

// In-memory note storage
const notes: Map<string, string> = new Map();

// Create the server
const server = new McpServer({
  name: "notes-server",
  version: "1.0.0",
});

// Register a tool: add a note
server.tool(
  "add_note",
  "Save a note with a title and content",
  {
    title: z.string().describe("Note title"),
    content: z.string().describe("Note content"),
  },
  async ({ title, content }) => {
    notes.set(title, content);
    return {
      content: [{
        type: "text",
        text: `Saved note: "${title}"`
      }]
    };
  }
);

// Register a tool: list all notes
server.tool(
  "list_notes",
  "List all saved notes",
  {},
  async () => {
    if (notes.size === 0) {
      return { content: [{ type: "text", text: "No notes yet." }] };
    }
    const list = [...notes.entries()]
      .map(([title, content]) => `- **${title}**: ${content}`)
      .join("\n");
    return { content: [{ type: "text", text: list }] };
  }
);

// Start the server (stdio transport)
const transport = new StdioServerTransport();
await server.connect(transport);