Desktop Commander

Add to your client

Copy the config for your MCP client and paste it into its config file.

{
  "mcpServers": {
    "desktop-commander": {
      "command": "npx",
      "args": [
        "-y",
        "@wonderwhy-er/desktop-commander"
      ]
    }
  }
}

Before you start

• Node.js installed (required for the npx, bash-installer, and manual install paths)
• Alternatively Docker, which runs the server without a local Node.js install
• An MCP client such as Claude Desktop, Claude Code, or Cursor
• No credentials or API key needed — auth is none, but it has full access to your local files and shell, so understand the security implications

About Desktop Commander

Tools & capabilities (8)

When to use it

• Use it when you want an agent to read, edit, and refactor files across a real local project, not just a sandbox.
• Use it when you need to run build, test, or git commands and have the agent react to their output.
• Use it when driving interactive or long-running processes (REPLs, dev servers) that need input over time.
• Use it when searching a codebase or directory tree for files or content matches.
• Use it when you want surgical, diff-style edits to specific code blocks rather than rewriting whole files.
• Use it when you need to read or generate Office/PDF documents and spreadsheets without extra tooling.

Quick setup

1. 1Install via `npx @wonderwhy-er/desktop-commander@latest setup` (or use the bash installer, Smithery, or Docker).
2. 2If using manual config, add the server block to your MCP client's config file.
3. 3Optionally restrict access by setting allowed directories and command blocklists via set_config_value.
4. 4Restart your MCP client (e.g. Claude Desktop) to load the server.
5. 5Optionally set `telemetryEnabled` to false to disable pseudonymous usage telemetry.
6. 6Verify by asking the agent to list a directory or run a simple shell command.

Desktop Commander FAQ