Getting Started with mcp-framework

Build your first MCP server in minutes using mcp-framework. Step-by-step guide from installation to connecting with Claude Desktop.

title: "Getting Started with mcp-framework" description: "Build your first MCP server in minutes using mcp-framework. Step-by-step guide from installation to connecting with Claude Desktop." keywords: ["MCP getting started", "mcp-framework tutorial", "build MCP server", "Claude Desktop", "TypeScript MCP"] date: "2025-01-15" updated: "2025-04-01" author: "QuantGeekDev" order: 5 level: "beginner" duration: "15 min" language: "en" topic: "getting-started"

Prerequisites

Before you begin, make sure you have:

• Node.js 18+ installed
• npm or another package manager
• A text editor (VS Code recommended)
• Claude Desktop (optional, for testing)

Step 1: Install mcp-framework

Install the framework globally to get access to the mcp CLI:

npm install -g mcp-framework

This gives you the mcp command, which you will use to create and manage MCP server projects.

Step 2: Create Your Project

Scaffold a new MCP server project:

mcp create my-mcp-server
cd my-mcp-server

This creates a ready-to-use project structure:

my-mcp-server/
  src/
    tools/
      ExampleTool.ts
    resources/
    prompts/
  package.json
  tsconfig.json

The framework uses auto-discovery — any tool, resource, or prompt you place in the corresponding directory is automatically registered with the MCP server.

Step 3: Build Your First Tool

Open src/tools/ExampleTool.ts to see a working tool:

import { MCPTool } from "mcp-framework";
import { z } from "zod";

class ExampleTool extends MCPTool<typeof inputSchema> {
  name = "example_tool";
  description = "An example tool that greets users";

  schema = {
    input: z.object({
      name: z.string().describe("The name to greet"),
    }),
  };

  async execute({ name }: z.infer<typeof this.schema.input>) {
    return `Hello, ${name}! Welcome to MCP.`;
  }
}

export default ExampleTool;

Each tool has:

• A name — how the AI refers to it
• A description — helps the AI understand when to use it
• A schema — defines the input parameters with Zod validation
• An execute method — the actual logic

Step 4: Build and Run

Compile your TypeScript and start the server:

npm run build

Your MCP server is now ready to accept connections.

Step 5: Connect to Claude Desktop

Add your server to Claude Desktop's configuration. Open claude_desktop_config.json:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json

Add your server:

{
  "mcpServers": {
    "my-mcp-server": {
      "command": "node",
      "args": ["/absolute/path/to/my-mcp-server/dist/index.js"]
    }
  }
}

Restart Claude Desktop, and your tool will be available in conversations.

Step 6: Add More Capabilities

Adding a Resource

Create src/resources/GreetingResource.ts:

import { MCPResource } from "mcp-framework";

class GreetingResource extends MCPResource {
  uri = "greeting://welcome";
  name = "Welcome Message";
  description = "A welcome message resource";
  mimeType = "text/plain";

  async read() {
    return {
      contents: [
        {
          uri: this.uri,
          text: "Welcome to my MCP server!",
          mimeType: this.mimeType,
        },
      ],
    };
  }
}

export default GreetingResource;

Adding a Prompt

Create src/prompts/AnalyzePrompt.ts:

import { MCPPrompt } from "mcp-framework";

class AnalyzePrompt extends MCPPrompt {
  name = "analyze";
  description = "Analyze a piece of text";
  arguments = [
    {
      name: "text",
      description: "The text to analyze",
      required: true,
    },
  ];

  async getMessages({ text }: { text: string }) {
    return [
      {
        role: "user" as const,
        content: {
          type: "text" as const,
          text: `Please analyze the following text:\n\n${text}`,
        },
      },
    ];
  }
}

export default AnalyzePrompt;

What You Built

You now have a complete MCP server with:

• A tool that Claude can call during conversations
• A resource that provides structured data
• A prompt template for common tasks
• Auto-discovery — just add files and rebuild

Next Steps

• Read the What is MCP? guide for deeper understanding
• Explore mcp-framework on GitHub for advanced features
• Check the official MCP documentation for protocol details
• View the npm package for latest releases

This guide is part of MCP International — MCP tutorials in every language. Created by @QuantGeekDev, author of mcp-framework (3.3M+ npm downloads, validated by Anthropic).