Atu Desktop Control

• Outcome: Review
• Latest audit: May 29, 2026, 9:28 PM
• Version:
  1.0.0

Overview

This skill appears to be a legitimate desktop automation tool, but it gives an agent broad screen, keyboard, clipboard, and window-control powers without enough documented user safeguards.

Install only if you intentionally want an agent to control your live desktop. Avoid using it while passwords, tokens, private chats, financial data, or sensitive documents are visible or on the clipboard, and require explicit confirmation before screenshots, clipboard reads, keyboard shortcuts, or closing windows.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (6)

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The README promotes full desktop automation capabilities including screenshots, keyboard input, clipboard access, and window control, but does not warn users about the privacy and system-integrity risks those features create. In an agent skill context, these capabilities can expose sensitive data, trigger unintended actions, or enable misuse of the user's desktop environment if operators are not clearly informed and protective boundaries are not documented.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: Clipboard contents often contain passwords, tokens, personal data, or proprietary text, and the skill exposes both read and write operations without any warning or consent guidance. In a desktop automation skill, clipboard access is especially sensitive because it can silently exfiltrate or overwrite user data during normal workflows.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The examples include actions like Alt+F4, window close, and broad keyboard/window manipulation, which can terminate applications, disrupt user work, or trigger unintended UI flows. Because this skill is explicitly designed for full desktop control, unsafe examples without cautionary guidance materially increase the chance of harmful misuse or accidental damage.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: Screenshots can capture emails, chats, credentials, tokens, documents, and other sensitive on-screen information, yet the documentation presents this capability without any privacy warning. In a desktop-control skill, screenshotting is inherently high-risk because it provides broad visibility into whatever is currently displayed on the user's screen.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The clipboard commands allow unrestricted reading and writing of system clipboard contents, which may contain passwords, tokens, API keys, personal data, or other sensitive material. In a desktop automation skill, this is especially dangerous because an agent can silently exfiltrate clipboard data or overwrite it to influence user actions without any warning, consent gate, or audit trail.

Missing User Warnings

Medium

• Confidence: 86% confidence
• Finding: The window close operation can terminate arbitrary windows based on a loose title-substring or hwnd match with no confirmation, which can cause data loss if unsaved work is present. In this skill context, the danger is elevated because the tool is designed for broad desktop control, so an agent could disrupt active applications or intentionally close security prompts, editors, or user sessions.

VirusTotal

55/55 vendors flagged this skill as clean.