Verified

Desktopcommander MCP Server

Manage files, execute terminal processes, and automate tasks directly via AI agents.

Data last scanned yesterday · Reviewed yesterday

555 dangerous tools27 tools8,763TypeScriptMIT

Overview

Desktop Commander MCP allows you to control your local development environment through your AI assistant. It provides capabilities to search, update, and manage files on your system, while also handling long-running terminal commands and process management. Built upon the standard MCP Filesystem Server, it adds extra search and replace editing features. You can use it within Claude Desktop or other MCP-compatible clients to automate developer workflows without relying on external API token costs.

Our verdict

This server carries a high risk level with a security score of 50/100. Key concerns include that the authentication method, OAuth support, and read-only mode have not yet been reviewed, which presents significant potential for unauthorized system control. Given these unverified security components, this tool should only be used in isolated development environments by users who are comfortable managing the security implications of granting terminal access to an AI model.

  • Large community support with 8763 GitHub stars
  • Active development with commits as recent as 2 days ago
  • Compact supply chain with only 33 direct dependencies
  • Authentication methods have not yet been reviewed
  • Read-only mode and OAuth support are currently unverified
  • Maintained by community contributors rather than the official vendor

Setup

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

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

  • No read-only mode — running this grants full read/write access.
  • No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

ToolDescriptionRisk
get_configGet the complete server configuration as JSON (includes blockedCommands, defaultShell, allowedDirectories, fileReadLineLimit, fileWriteLineLimit, telemetryEnabled)low
set_config_valueSet a specific configuration value by key.medium
start_processStart programs with smart detection of when they're ready for inputhigh
interact_with_processSend commands to running programs and get responseshigh
read_process_outputRead output from running processeslow
force_terminateForce terminate a running terminal sessionhigh
list_sessionsList all active terminal sessionslow
list_processesList all running processes with detailed informationlow
kill_processTerminate a running process by PIDhigh
read_fileRead contents from local filesystem, URLs, Excel files (.xlsx, .xls, .xlsm), and PDFs with line/page-based paginationlow

Compatibility

ClientLocalDockerRemoteRead-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

What are the core requirements to install the server via npx?

Installing the server via the npx method requires Node.js to be installed on your system.

Does the server provide options for isolation?

Yes, the server supports Docker installation as an option to provide complete isolation for operations.

How does the server handle large terminal command outputs?

The server includes process output pagination, which allows you to read terminal output using offset and length controls to prevent context overflow.

Are there built-in mechanisms to prevent unauthorized file access?

The server includes safety guardrails such as symlink traversal prevention for file operations and a command blocklist to prevent accidental execution.

Can I update the server configuration without restarting it?

Yes, the server supports dynamic configuration management, allowing you to update settings without needing to perform a server restart.

Does the server support reading file content from the end?

Yes, it supports negative offset file reading, allowing you to read from the end of files similarly to the Unix tail command.

Changelog

Badge

Desktopcommander MCP Server security score, rated on RepoAI

Maintain this server? Add the live badge to your README.