How to use DesktopCommanderMCP: install, setup and tips

What Desktop Commander does

Desktop Commander is an MCP (Model Context Protocol) server. It gives an AI assistant controlled access to your local machine. It is built on top of the MCP Filesystem Server and adds search and replace file editing on top of it.

The project's README lists capabilities including executing terminal commands with output streaming, process management (list and kill processes), searching files and content, and surgical text replacements for small changes or full file rewrites for major changes.

• Run terminal commands and manage long-running processes such as dev servers or database sessions
• Search folders and file content with vscode-ripgrep based recursive search
• Read and edit text, Excel (.xlsx, .xls, .xlsm), PDF and DOCX files
• Read the end of a file with negative offsets, similar to Unix tail
• Keep a local record of tool calls on the machine running it

Install Desktop Commander for Claude Desktop

The simplest route is the npx setup command from the README. The prerequisite is Node.js. Open a terminal and run the command below. It installs and configures the server for Claude Desktop.

After it finishes, restart Claude if it is running. The README notes that this install method updates automatically when you restart Claude.

To remove it, the README gives this command. You can also remove the entry from your claude_desktop_config.json file.

• Config file on Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
• Config file on Windows: %APPDATA%\Claude\claude_desktop_config.json
• Config file on Linux: ~/.config/Claude/claude_desktop_config.json
• Setup option --no-onboarding disables onboarding prompts for new users
• Setup option --debug enables Node.js inspector debugging

Other ways to install it

The README lists several other options. A bash script installer is provided for macOS and installs Node.js if needed. Smithery offers an install with a provided key after you choose Claude Desktop. You can also add a JSON entry to claude_desktop_config.json by hand, or clone the repository and run its setup script.

Manual cloning is the only option the README marks as requiring manual updates. Docker is also listed for users who do not have Node.js installed, and it runs in a container.

• npx setup: requires Node.js, auto-updates
• macOS bash installer: installs Node.js if needed, auto-updates
• Smithery: requires Node.js, auto-updates
• Manual config entry: requires Node.js, auto-updates
• Git clone: requires Node.js, manual updates
• Docker: no Node.js required, auto-updates

Use it safely: understand the guardrails

The README states plainly that Desktop Commander is not a sandbox. It includes safety guardrails: symlink traversal prevention on file operations, a command blocklist for accidental execution, and optional Docker isolation. Read SECURITY.md in the repository before you grant access to important folders.

Treat the assistant as having access to your files and shell. Start with a project folder rather than your whole home directory, and review commands it proposes.

• Guardrails reduce accidents; they are not a security boundary
• The command blocklist is for accidental execution
• Docker isolation is available if you want a separated environment

Edit and search files with Claude

Ask Claude in plain language to find text, change a function, or read a spreadsheet. The README describes two editing modes: surgical text replacements for small changes, and full file rewrites for major changes. Pattern-based replacements and multi-file edits are also listed.

For long command output, the README mentions process output pagination with offset and length controls, which helps avoid overflowing the assistant's context.