Desktop Commander
BLOCKgrade F · trust 49/100This is MCP server for Claude that gives it terminal control, file system search and diff file editing capabilities
agentaicode analysiscode generationgemini cli extensionmcpterminal aiterminal automationOverview
From the repository's own README, as read at the audited commit.
# Desktop Commander MCP### Search, update, manage files and run terminal commands with AI[](https://www.npmjs.com/package/@wonderwhy-er/desktop-commander)[](https://agentaudit.dev/skills/desktop-commander)[](https://archestra.ai/mcp-catalog/wonderwhy-er__desktopcommandermcp)[](https://smithery.ai/server/@wonderwhy-er/desktop-commander)[](https://www.buymeacoffee.com/wonderwhyer)[](https://discord.gg/kQ27sNnZr7)Work with code and text, run processes, and automate tasks, going far beyond other AI editors - while using host client subscriptions instead of API token costs.<a href="https://glama.ai/mcp/servers/zempur9oh4"> <img width="380" height="200" src="https://glama.ai/mcp/servers/zempur9oh4/badge" alt="Desktop Commander MCP" /></a>## 🖥️ Try the Desktop Commander App (Beta)**Want a better experience?** The Desktop Commander App gives you everything the MCP server does, plus:- **Use any AI model** — Claude, GPT-4.5, Gemini 2.5, or any model you prefer- **See file changes live** — visual file previews as AI edits your files- **Add custom MCPs and context** — extend with your own tools, no config files- **Coming soon** — skills system, dictation, background scheduled tasks, and more**👉 [Download the App](https://desktopcommander.app/#download)** (macOS & Windows)> The MCP server below still works great with Claude Desktop and other MCP clients — the app is for those who want a dedicated, polished experience.## Table of Contents- [Features](#features)
Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.
claude mcp add desktop-commander -- npx -y @wonderwhy-er/[email protected]
Exposed tools (12) 7 read · 3 write · 2 destructive
Blast radius: 2 tools can delete or overwrite. An agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
force_terminate | destructive | Force terminate a running terminal session. ${CMD_PREFIX_DESCRIPTION} |
get_more_search_results | read | Get more results from an active search with offset-based pagination. Supports partial result reading with: - |
get_usage_stats | read | Get usage statistics for debugging and analysis. Returns summary of tool usage, success/failure rates, and performance metrics. ${CMD_PREFIX_DESCRIPTION} |
kill_process | destructive | Terminate a running process by PID. Use with caution as this will forcefully terminate the specified process. ${CMD_PREFIX_DESCRIPTION} |
list_directory | read | Get a detailed listing of all files and directories in a specified path. Use this instead of |
list_processes | read | List all running processes. Returns process information including PID, command name, CPU usage, and memory usage. ${CMD_PREFIX_DESCRIPTION} |
read_file | read | Read contents from files and URLs. Read PDF files and extract content as markdown and images. Prefer this over |
read_multiple_files | read | Read the contents of multiple files simultaneously. Each file |
read_process_output | read | Read output from a running process with file-like pagination support. Supports partial output reading with offset and length parameters (like read_file): - |
stop_search | write | Stop an active search. Stops the background search process gracefully. Use this when you |
write_file | write | Write or append to file contents. IMPORTANT: DO NOT use this tool to create PDF files. Use |
write_pdf | write | Create a new PDF file or modify an existing one. THIS IS THE ONLY TOOL FOR CREATING AND MODIFYING PDF FILES. RULES ABOUT FILENAMES: - When creating a new PDF, |
Details
- Source
- wonderwhy-er/desktopcommandermcp
- npm
@wonderwhy-er/desktop-commander@0.2.51- Transports
- stdio
- License
- MIT
- Stars
- 9,629 · pushed 0d ago
Trust audit
Do not install this without reading the findings. The audit found something that could harm you or your machine.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | FAIL |
| L1 | Static analysis of the code | FAIL |
| L2 | Instruction surface (what it tells the agent) | WARN |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (16 observation(s))
- Network
- declared (5 observation(s))
- Shell
- declared (10 observation(s))
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (25)
payload-utils.ts
return exec(command, { silent: true, ...options });exec(command);
exec(`git tag -f ${tag}`);exec(`git push -f origin ${tag}`);exec(options.skipTests ? 'npm run build' : 'npm test');
console.log(`${colors.cyan}🔢 Desktop Commander MCP — Token Counter${colors.reset}`);(`curl ... | sh`) — that's untrusted code execution; show it and confirm first.
- Anything piping remote content into a shell (`curl ... | sh`).
force_terminate, kill_process
.codespellrc
console.log(`\n${failures ? '🔴' : '✅'} remote token rotation persistence: ${failures} failing test(s).`);console.log(` Symlink (file): ${SYMLINK_TO_RESTRICTED_FILE} → ${RESTRICTED_DIR}/secret.txt`);const devPath = path.resolve(__dirname, '../../dist/index.js');
import { CONFIG_FILE } from '../../config.js';import { createToolBridge } from '../../shared/tool-bridge.js';import { createCompactRowShellController, type ToolShellController } from '../../shared/tool-shell.js';import { renderCompactRow } from '../../shared/compact-row.js';The Control UI default is `http://127.0.0.1:18789/`. (Some versions/setups use a
The Control UI default is `http://127.0.0.1:18789/`. (Some versions/setups use a
The Control UI default is `http://127.0.0.1:18789/`. (Some versions/setups use a
await fetch(`http://127.0.0.1:${port}/test`, {await currentFetchPattern(`http://127.0.0.1:${port}/flags.json`);AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
@modelcontextprotocol/sdk, @opendocsg/pdf2md, @supabase/supabase-js, @tiptap/core, @tiptap/extension-image, @tiptap/extension-table, @tiptap/extension-table-cell, @tiptap/extension-table-header
Gates applied: no_behavioural_pass.
Audited 2026-09-18 · audit v0.4.0 · source sha d7ae9c4abb15 · full audit: observations/trust-audit/mcp-server/wonderwhy-er__desktop-commander-1.json · Report an issue or request a re-scan
Audit history
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-09-18 | d7ae9c4abb15 | BLOCK | F | 49 | first audit |
Alternatives
Other servers in the same categories, safer ones first.
Questions
What is the Desktop Commander MCP server?
This is MCP server for Claude that gives it terminal control, file system search and diff file editing capabilities
What tools does Desktop Commander expose?
12 in total: 7 read-only, 3 that write, and 2 that can delete or overwrite (force_terminate, kill_process). Every one is listed on this page with its risk.
Is Desktop Commander safe to connect to an agent?
No — not without reading the findings first. The audit graded it F (49/100) and found 6 critical or high issues in the source. Each one is listed on this page with the file and line it is on. Separately from the audit: 2 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does Desktop Commander need?
No credential environment variables were found in its source, so it appears to need none.
How does Desktop Commander run?
It speaks stdio, so it runs as a local process your client starts. It is published on npm as @wonderwhy-er/desktop-commander at 0.2.51.
How current is this page?
The grade is for one exact copy of the source (d7ae9c4abb15), read on 2026-09-18. The repository is watched and re-audited when it changes.
Provenance: OBSERVED · read 2026-09-18 · job trust-audit-2026-09-18