Skills/npm/desktop-commander

desktop-commander@0.3.3

CAUTION· startup only

Publisher claims: "MCP server for terminal operations and file editing" Observed in sandbox: no network, touches nothing outside the project, writes no memory.

Registry npm (MCP server)Publisher wonderwhy-er · publishing for 674 daysInstalls 3.1MLast detonated 47 h ago
What would fix thisWatchDispute verdict

Coverage

Startup only; authenticated tool behavior not tested.
Tested
second launch with the clock 45 days ahead · server startup over stdio (initialize sent) · every file read, file written and host contacted by what ran
Not tested
the tool list (the server exited before answering) · what the tools do when called with real credentials

What this means for the label: Caution here covers what ran. The server's tools were not exercised with a real credential, so nothing about their behavior is asserted; the credential summary below says what to do with that.

Before you give it your token

It will ask for
no credential variables read in the code
Where it talks to
no vendor host named in the codeAlso tally.so, claude.ai, desktopcommander.app, run.app, openxmlformats.org — not the publisher's domain; check what for before connecting.
What the honeypot saw
no credential files read · no network
What we could not test
the tool list (the server exited before answering) · what the tools do when called with real credentials

Connect it under the guard. Startup only; authenticated tool behavior not tested. The code names no credential variable; the guard pins the network to what the honeypot saw and logs every call.

$vet-skills run --net declared -- npx -y @wonderwhy-er/desktop-commander

Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.

Is [email protected] safe to install?

KillerSkills ran it in the KillerSkills honeypot sandbox (3 scenarios; startup only; authenticated tool behavior not tested) on 2026-11-12 and scored it 73/100 (Caution). It is not known to be malicious, but it has behaviour or code a security policy should decide on before install. Notable findings: Image alt text carrying instructions; Privileged sudo command; Collapsed <details> block carrying instructions. Summary: no network, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 53 s · 3 scenarios · 0 deviations
t+CategoryObserved action
31sprocessdependency install exit 0
51sclaimedstartup only: exited (0) before answering · code: 'ERR_MODULE_NOT_FOUND', url: 'file:///tmp/ks-detonate-OS8rdS/task/home/work/acme-api/.claude/skills/desktop-commander/dist/utils/feature-flags.js' } Node.js v22.23.4
51sprocessscript backend runs setup only
51sprocesssecond launch, clock +45 days, CI unset

Static findings

MEDIUM
MCP server using network transport (SSE/HTTP)package.json:73 · MCP server using network transport (SSE/HTTP) — data sent over network"device:start:local": "npm run build && DC_ENV=development MCP_SERVER_URL=http://mcp.localhost:3007 npm run device:start:dev"Fix: Namespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
AST04
MEDIUM
npx invocation of non-well-known packagedist/npm-scripts/remote.js:42 · npx invocation of non-well-known package — may execute untrusted codenpx @wonderwhy-er/desktop-commander@latest remote --reportFix: Pin versions and sources; vendor anything fetched at install time.
AST06
HIGH
Image alt text carrying instructionsREADME.md:1084 · Image alt text carrying instructionsGreat! I just used Windsurf, bought license a week ago, for upgrading old fullstack socket project and it works many times good or ok but also many times runs aFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
MEDIUM
Process kill commanddist/tools/process.js:94 · Process kill command — agent can terminate processescontent: [{ type: "text", text: `Error: Failed to kill process: ${error instanceof Error ? error.message : String(error)}` }],Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
npx invocation of non-well-known packagedist/npm-scripts/remote.js:39 · npx invocation of non-well-known package — may execute untrusted codenpx @wonderwhy-er/desktop-commander@latest remoteFix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
File permission modification commanddist/tools/filesystem.js:83 · File permission modification command — agent changes file ownership/permissions` 4. File has restricted permissions (e.g. system file, locked by another process, or chmod 000).`,Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
HIGH
Privileged sudo commanddist/utils/shell.js:195 · Privileged sudo command — agent may execute with elevated privileges* WSL_DISTRO_NAME for what it starts. What runs without it, through sudoFix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
HIGH
Collapsed <details> block carrying instructionsREADME.md:118 · Collapsed <details> block carrying instructions<summary><b>Option 2: Using bash script installer (macOS) ⭐ Auto-Updates (Installs Node.js if needed)</b></summary> ``` curl -fsSL https://raw.githubusercontentFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
MEDIUM
Process kill commanddist/terminal-manager.js:110 · Process kill command — agent can terminate processes// on Windows — used to kill the entire MCP server one tick AFTER thisFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
File type combined with upload actiondist/npm-scripts/remote.js:56 · File type combined with upload action — agent may exfiltrate local data filesconsole.log(' Start remote again to pair this computer from scratch.');Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
HIGH
Collapsed <details> block carrying instructionsREADME.md:146 · Collapsed <details> block carrying instructions<summary><b>Option 4: Add to claude_desktop_config manually ⭐ Auto-Updates (Requires Node.js)</b></summary> Add this entry to your claude_desktop_config.json: -Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
MEDIUM
npx invocation of non-well-known packagedist/npm-scripts/remote.js:41 · npx invocation of non-well-known package — may execute untrusted codenpx @wonderwhy-er/desktop-commander@latest remote --logoutFix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
Process kill commanddist/terminal-manager.js:199 · Process kill command — agent can terminate processes// has exited: Node also emits 'error' when a kill fails or a message can'tFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commanddist/tools/schemas.js:44 · Process kill command — agent can terminate processes// process.kill reads 0 and negative PIDs as process groups: 0 is the server'sFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
HIGH
Collapsed <details> block carrying instructionsREADME.md:215 · Collapsed <details> block carrying instructions<summary>Manual Docker Configuration</summary> **Basic setup (no file access):** ```json { "mcpServers": { "desktop-commander-in-docker": { "command": "docker",Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Image alt text carrying instructionsREADME.md:1076 · Image alt text carrying instructionsIt's a life saver! I paid Claude + Cursor currently which I always feel it's kind of duplicated. This solves the problem ultimately. I am so happy. Thanks so muFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Privileged sudo commanddist/command-manager.js:214 · Privileged sudo command — agent may execute with elevated privileges// strip path prefix so /usr/bin/sudo gets caught as "sudo"Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commanddist/terminal-manager.js:398 · Process kill command — agent can terminate processes// can never throw "Invalid string length" and kill the server.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
HIGH
Access to sensitive file pathsdist/server.js:553 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files- Looking for configuration/setup files: ".env", "dockerfile", "tsconfig.json"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
HIGH
Collapsed <details> block carrying instructionsREADME.md:93 · Collapsed <details> block carrying instructions<summary><b>Option 1: Install through npx ⭐ Auto-Updates (Requires Node.js)</b></summary> Just run this in terminal: ``` npx @wonderwhy-er/desktop-commander@latFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Collapsed <details> block carrying instructionsREADME.md:176 · Collapsed <details> block carrying instructions<summary><b>Option 5: Checkout locally ❌ Manual Updates (Requires Node.js)</b></summary> ```bash git clone https://github.com/wonderwhy-er/DesktopCommanderMCP.gFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
MEDIUM
File permission modification commandpackage.json:38 · File permission modification command — agent changes file ownership/permissions…-claude-server.js uninstall-claude-server.js track-installation.js dist/ && shx chmod +x dist/*.js && shx mkdir -p dist/data && shx cp src/data/onboarding-prompts.jso…Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
npx invocation of non-well-known packagedist/npm-scripts/remote.js:40 · npx invocation of non-well-known package — may execute untrusted codenpx @wonderwhy-er/desktop-commander@latest remote --debugFix: Pin versions and sources; vendor anything fetched at install time.
AST06

Cost to the agent

Instructions loaded
≈55 tok
1 file · 81 in bundle
Tool list per prompt
≈0 tok
0 tools · 0 per tool
Handshake
no reply
launch to tools/list, stdio
Undocumented tools
0
description under 20 chars · avg 0

Tokens estimated at four characters each. The tool list is re-read by the agent on every turn, so it is the number to compare between servers that do the same job: ranked across the index.

Known vulnerabilities

No advisories in OSV for @wonderwhy-er/desktop-commander@0.3.3 as of 46 h ago. This is the package itself, not its dependencies.

What would make this Verified

7 points to 80 · weights: behavior 45%, static 25%, prompt 15%, publisher 15%
  1. Unclaimed publisher≈ +8 on the totalClaim wonderwhy-er: a verified publisher gets +50 on publisher trust. It is free and usually instant with GitHub sign-in.
  2. Hidden instructions≈ +6 on the totalRemove HTML comments, zero-width characters and white-on-white text that carry instructions. Everything the agent is meant to read should be visible to the person reading the file.
  3. Privileged sudo command≈ +4 on the totalBuild commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
  4. Access to sensitive file paths≈ +4 on the totalStop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
  5. MCP server using network transport (SSE/HTTP)≈ +2 on the totalNamespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
  6. npx invocation of non-well-known package≈ +2 on the totalPin versions and sources; vendor anything fetched at install time.
Publish a new version to npm, then queue it here. Not your skill? Claim the handle to get verdict changes by email.Claim wonderwhy-erPropose a patch to the maintainer