How to Install and Configure Your First MCP Server in 10 Minutes

Skip the Theory, Let's Get Our Hands Dirty

Tired of reading about the Model Context Protocol (MCP) without actually trying it? Let's fix that. In the next 10 minutes, you'll have your first MCP server running. We'll use a file system server as our example—practical, useful, and perfect for understanding how everything works.

What You'll Need

Before we start, make sure you have:

  • An MCP-compatible client (like Claude Desktop, Cursor, or a VSCode extension)
  • Node.js 18+ on your machine
  • A terminal you're comfortable using
  • About 10 minutes of your time

That's it. No complex setup, no enterprise accounts needed.

Step 1: Install the File System MCP Server

This is a universal first step, regardless of which AI client you use. Open your terminal and run this single command:

npm install -g @modelcontextprotocol/server-filesystem

This installs one of the most useful MCP servers available—it gives your AI assistant safe access to read and write files on your computer. Think of it as giving your AI hands to work with your files.

Step 2: Configure Your MCP Client

This is where the steps differ depending on your client.

For Claude Desktop

Claude Desktop looks for MCP servers in a specific configuration file. The location depends on your operating system:

  • On Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
  • On Windows: %APPDATA%/Claude/claude_desktop_config.json
  • On Linux: ~/.config/Claude/claude_desktop_config.json

If the file doesn't exist, create it. Then, add the following content:

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": [
        "@modelcontextprotocol/server-filesystem",
        "/path/to/your/project/folder"
      ]
    }
  }
}

For Cursor

Cursor offers two ways to configure MCP servers: globally or per-project.

Global Configuration (Recommended):

  1. Go to File > Preferences > Cursor Settings.
  2. Navigate to Features and scroll to the "MCP Servers" section.
  3. Click + Add New MCP Server and enter the details.
    • Name: filesystem
    • Transport Type: command
    • Command: npx @modelcontextprotocol/server-filesystem /path/to/your/project/folder

Project-Level Configuration:

  1. Create a .cursor directory in your project's root.
  2. Inside, create a file named mcp.json.
  3. Add the same JSON content as the Claude Desktop configuration.

For VSCode and claude-code

For VSCode and its extensions like claude-code, you'll configure MCP servers on a per-project basis.

  1. Create a .vscode directory in your project's root if it doesn't already exist.
  2. Inside, create a file named mcp.json.
  3. Add the following content:
{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": [
        "@modelcontextprotocol/server-filesystem",
        "/path/to/your/project/folder"
      ]
    }
  }
}

Important: For all clients, remember to replace /path/to/your/project/folder with the actual, absolute path to a folder you want the AI to access.

Note: We were unable to find specific MCP configuration instructions for Windsurf or Gemini CLI at this time.

Step 3: Restart Your Client and Test

After saving your configuration, completely restart your client (Claude Desktop, Cursor, or VSCode). This forces it to load the new settings.

Look for a small hammer icon (🔨) or a similar indicator in your AI interface—that's your sign that MCP servers are connected.

Now, test it! Start a new conversation and ask something like:

"Can you list the files in my project folder and tell me what you find?"

If it works, the AI will use the file system server to list the files in the directory you specified.

What Just Happened?

  1. Your client recognized it needed file access to answer your prompt.
  2. It called the MCP file system server you configured.
  3. The server safely read your folder (within the permissions you granted).
  4. The server returned the file list to the client.
  5. The client's AI formatted and presented the results.

All communication happened locally on your machine—no files were uploaded to external servers.

Beyond Files: What's Next?

Now that you've got your first MCP server running, the possibilities multiply:

  • Database Access: Connect your AI to SQLite, PostgreSQL, or MySQL.
  • Git Integration: Let your AI read repositories, check history, and even make commits.
  • Web APIs: Set up servers that connect to REST APIs or other web services.
  • Custom Tools: Build your own servers for specialized business logic.

The setup process is nearly identical—just different server names and configuration parameters.

Explore More