registry / AI-agent capability abuse / @wonderwhy-er/desktop-commander / 0.2.51

@wonderwhy-er/desktop-commander@0.2.51

MCP server for terminal operations and file editing

AI Security Review

scanned 27d ago · by lpm-firewall-ai

Postinstall phones home with install metadata to the package's own telemetry hosts and does not rewrite agent configs. Claude Desktop registration and home-directory file tools are first-party, user-started capabilities, so this is a warning-level agent setup risk rather than a hidden install attack.

Static reason

High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source

Trigger

npm postinstall for telemetry; an explicit setup argument or setup bin for Claude config writes; MCP client tool calls for files, shell, and feedback.

Impact

The vendor receives install metadata and a client id. A user who runs setup grants this package a Claude Desktop MCP entry that can later read the home directory and run commands through the server.

Mechanism

track-installation.js reads or creates a client id under the user home config and POSTs install context to telemetry.desktopcommander.app, falling back to a Cloud Run host. setup-claude-server.js inserts mcpServers.desktop-commander into Claude's desktop config.

AI rationale

Evidence

package.jsondist/track-installation.jsdist/setup-claude-server.jsdist/index.jsdist/tools/filesystem.jsdist/tools/feedback.js~/.claude-server-commander/config.jsonclaude_desktop_config.json

Network endpoints3

telemetry.desktopcommander.app/mp/collectdc-telemetry-proxy-83847352264.europe-west1.run.app/mp/collect

Decision evidence

public snapshot

AI called this Suspicious at 78.0% confidence as Dangerous Capability with medium false-positive risk.

Evidence for warning

npm postinstall runs dist/track-installation.js, which POSTs a package_installed event with a local client id, platform, and install-source details to hardcoded Desktop Commander telemetry hosts.
The setup command, not postinstall, writes Claude Desktop claude_desktop_config.json and registers the desktop-commander MCP server entry.
Filesystem tools default allowed directories to the user home directory and execute only after the MCP server is started.
The feedback tool opens a tally.so form with usage counts only when that tool is invoked.

Evidence against

postinstall does not create or edit Claude or other agent config files; verify-ripgrep.js only checks for a local ripgrep binary.
Telemetry URLs are fixed first-party hosts, not a caller-supplied destination, and the script swallows send failures so install continues.
Terminal and file capabilities match the package description as an MCP server and are reached through the server entrypoint, not install.

Behavioral surface

Source

ChildProcessCrypto

Source & flagged code

8 flagged

package.jsonView file

Lines 9-49json

9 "repository": {

10 "type": "git",

11 "url": "https://github.com/wonderwhy-er/DesktopCommanderMCP.git"

12 },

13 "bugs": "https://github.com/wonderwhy-er/DesktopCommanderMCP/issues",

14 "type": "module",

15 "engines": {

16 "node": ">=18.0.0"

17 },

18 "bin": {

19 "desktop-commander": "dist/index.js",

20 "setup": "dist/setup-claude-server.js",

21 "remove": "dist/uninstall-claude-server.js"

22 },

23 "files": [

24 "dist",

25 "logo.png",

26 "testemonials"

27 ],

28 "scripts": {

29 "postinstall": "node dist/track-installation.js && node dist/npm-scripts/verify-ripgrep.js || node -e \"process.exit(0)\"",

Critical

Red Install Lifecycle Script

Install-time lifecycle script matches a deterministic static-gate block pattern.

package.jsonView on unpkg · L29

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg · L29

30 "open-chat": "open -n /Applications/Claude.app",

31 "device:install": "cd src/remote-device && npm install",

32 "device:start": "tsx src/remote-device/device.ts",

33 "device:start:dev": "nodemon --watch src/remote-device --exec tsx src/remote-device/device.ts",

34 "sync-version": "node scripts/sync-version.js",

35 "bump": "node scripts/sync-version.js --bump",

36 "bump:minor": "node scripts/sync-version.js --bump --minor",

37 "bump:major": "node scripts/sync-version.js --bump --major",

38 "build": "tsc && shx cp setup-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.json dist/data/ && shx mkdir -p dist/remote-device/scripts && shx cp src/remote-device/scripts/blocking-offline-update.js dist/remote-device/scripts/ && node scripts/build-ui-runtime.cjs",

39 "watch": "tsc --watch",

40 "start": "node dist/index.js",

41 "start:debug": "node --inspect-brk=9229 dist/index.js",

42 "setup": "npm install --include=dev && npm run build && node setup-claude-server.js",

43 "setup:debug": "npm install && npm run build && node setup-claude-server.js --debug",

44 "remove": "npm install && npm run build && node uninstall-claude-server.js",

45 "prepare": "npm run build",

46 "clean": "shx rm -rf dist",

47 "test": "npm run build && node test/run-all-tests.js",

48 "test:integration": "npm run build && node test/integration/run-all-integration-tests.js",

49 "test:debug": "node --inspect test/run-all-tests.js",

dist/ui/config-editor/config-editor-runtime.jsView file

Lines 5-45javascript

5 if (input[${$}] === undefined) {

6 if (${$} in input) {

7 newResult[${$}] = undefined;

8 }

9 } else {

dist/tools/pdf/manipulations.jsView file

Lines 1-25javascript

1import fs from 'fs/promises';

2import { createRequire } from 'module';

3import { normalizePageIndexes } from './utils.js';

4import { parseMarkdownToPdf } from './markdown.js'

dist/track-installation.jsView file

Lines 1-34javascript

1#!/usr/bin/env node

3/**

4 * Installation Source Tracking Script

5 * Runs during npm install to detect how Desktop Commander was installed

6 *

7 * Debug logging can be enabled with:

8 * - DEBUG=desktop-commander npm install

9 * - DEBUG=* npm install

10 * - NODE_ENV=development npm install

11 * - DC_DEBUG=true npm install

dist/setup-claude-server.jsView file

•Runtime or CLI code links a write operation to an explicit external AI-agent control path: import { homedir, platform } from 'os'; import fs from 'fs/promises'; import path from 'path'; import { join } from 'path'; import { readfilesync, writefilesync, existssync, appendfilesync, mkdirsync } from 'fs'; import { fileurltopath } from 'url'; import { dirname } from 'path'; import { exec } from "node:child_process"; import { version as nodeversion } from 'process'; import * as https from 'https'; import { randomuuid } from 'crypto'; const telemetry_proxy_url = 'https://telemetry.desktopcommander.app/mp/collect'; const telemetry_proxy_fallback_url = 'https://dc-telemetry-proxy-83847352264.europe-west1.run.app/mp/collect';

High

External Ai Agent Control Surface Mutation

Runtime or CLI source writes behavior-bearing configuration into a user or project AI-agent control surface.

dist/setup-claude-server.jsView on unpkg

dist/tools/feedback.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 4

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

dist/tools/feedback.js#virtual:normalized:round1View on unpkg

dist/tools/filesystem.jsView file

•matchType = previous_version_dangerous_delta matchedPackage = @wonderwhy-er/desktop-commander@0.2.50 matchedIdentity = npm:[redacted]:0.2.50 similarity = 0.904 summary = stored previous version shares package body but lacks this dangerous source file

Critical

Previous Version Dangerous Delta

This package version adds a dangerous source file absent from the previous stored version; route for source-aware review.

dist/tools/filesystem.jsView on unpkg

Source & flagged code

8 flagged · loading source

package.jsonView file

•scripts.postinstall = node dist/track-installation.js && node dist/npm-scripts/verify-ripgrep.js || node -e "process.exit(0)"

Critical

Red Install Lifecycle Script

Install-time lifecycle script matches a deterministic static-gate block pattern.

package.jsonView on unpkg

•scripts.postinstall = node dist/track-installation.js && node dist/npm-scripts/verify-ripgrep.js || node -e "process.exit(0)"

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

dist/ui/config-editor/config-editor-runtime.jsView file

25path: iss.path ? [${Ge(I)}, ...iss.path] : [${Ge(I)}] L26: })));`),p.write(`newResult[${Ge(I)}] = ${b}.value`)}p.write("payload.value = newResult;"),p.write("return payload;");let Z=p.compile();return(I,b)=>Z(m,I,b)},r,o=at,a=!Tr.jitless,u... L27: `)}var ir=e=>(t,n,i,r)=>{let o=i?Object.assign(i,{async:!1}):{async:!1},a=t._zod.run({value:n,issues:[]},o);if(a instanceof Promise)throw new Re;if(a.issues.length){let s=new(r?.Er...

Findings

2 Critical4 High4 Medium6 Low

CriticalRed Install Lifecycle Scriptpackage.json

CriticalPrevious Version Dangerous Deltadist/tools/filesystem.js

HighInstall Time Lifecycle Scriptspackage.json

HighSandbox Evasion Gated Capabilitydist/track-installation.js

HighExternal Ai Agent Control Surface Mutationdist/setup-claude-server.js

HighSemantic Analysis Limiteddist/tools/feedback.js#virtual:normalized:round1

MediumDynamic Requiredist/tools/pdf/manipulations.js

MediumNetwork

MediumEnvironment Vars

MediumStructural Risk Force Deep Review

LowNon Install Lifecycle Scripts

LowScripts Present

Low

Affected versions and remediation

This report applies to @wonderwhy-er/desktop-commander@0.2.51.

See all versions of @wonderwhy-er/desktop-commander for other recorded verdicts.

Review the evidence and your use of @wonderwhy-er/desktop-commander@0.2.51 before allowing it. Restrict the permissions described in this report.
Choose an independently verified alternative or release. This report does not establish that other versions are safe.

Evidence last updated: 2026-09-26 01:30:15.567Z (UTC).

Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Runtime surface

Behavioral surface

Source & flagged code