Win Mouse Native (Windows)

• Outcome: Pass
• Latest audit: May 27, 2026, 10:07 PM
• Version:
  1.0.0

Overview

This skill transparently provides Windows mouse movement and clicking through local scripts, with no evidence of exfiltration, persistence, or hidden behavior.

Install only if you want an agent to control your Windows mouse. Review the .cmd and .ps1 text before saving them as executable scripts, use explicit coordinates or verified targets, and supervise clicks or drags on payment pages, permission prompts, account settings, or other sensitive screens.

SkillSpector

By NVIDIA

Vulnerability Patterns

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

Findings (3)

Vague Triggers

Low

• Confidence: 83% confidence
• Finding: The metadata description says to use the skill whenever the user asks to move, click, drag, or automate pointer actions on Windows, which is a broad trigger for a capability that directly manipulates the host UI. Overbroad activation can cause the agent to invoke real mouse actions in situations where confirmation, tighter scoping, or safer alternatives would be more appropriate, increasing the chance of unintended system interaction.

Vague Triggers

Low

• Confidence: 89% confidence
• Finding: The instruction 'When the user asks to move/click the mouse' is open-ended and lacks constraints around context, confirmation, and prohibited targets. Because the skill controls the real system pointer, this ambiguity can let ordinary phrasing trigger UI automation against unintended applications or sensitive screens.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The OpenClaw usage section instructs the agent to execute mouse-control commands but does not explicitly warn that these actions affect the live Windows desktop and may click, drag, or alter state in whatever application is currently focused. Without a prominent warning and confirmation model, users or downstream agents may underestimate the side effects and accidentally trigger destructive or unsafe interactions.

VirusTotal

45/45 vendors flagged this skill as clean.