Security audit

Linux GUI Control

Security checks across malware telemetry and agentic risk

Overview

This skill gives an agent explicit Linux desktop-control tools, which are powerful but disclosed, purpose-aligned, and not paired with hidden collection or persistence.

Install only if you intentionally want an agent to operate your local Linux desktop. Before use, close or hide sensitive windows, verify the active target window, avoid visible passwords or private data during screenshots/UI inspection, and save work before any workflow that kills or restarts an application.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill enables screenshot capture, UI inspection, and synthetic keyboard/mouse input against arbitrary desktop applications without documenting privacy, consent, or safety boundaries. In a desktop environment, this can expose sensitive on-screen data and trigger unintended destructive actions in the wrong window or context.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The documented workflow uses `pkill` followed by relaunching an app with accessibility flags, but does not warn that this can terminate active sessions, discard unsaved work, or interrupt security-sensitive applications. In GUI automation contexts, abrupt process termination can cause data loss and service disruption even if the underlying intent is operational convenience.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.