Back to skill

Security audit

Mac System Tool Free

Security checks across malware telemetry and agentic risk

Overview

This macOS system-control skill is mostly purpose-aligned, but its broad/mismatched trigger instructions and sensitive local actions warrant careful review before installation.

Install only if you intentionally want an agent to run macOS shell commands for local system control. Keep prompts explicit, review any command before execution, require confirmation for screenshots, clipboard reads, process termination, sudo, and Trash emptying, and be aware that the skill's trigger text is too broad for its actual purpose.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Intent-Code Divergence

Medium
Confidence
83% confidence
Finding
The skill claims the free edition is focused on read-only queries and basic control, yet later documents a destructive command to recursively delete Trash contents. This inconsistency can cause users or downstream agents to grant more trust than warranted and may lead to unintended destructive execution under a misleading safety model.

Vague Triggers

High
Confidence
84% confidence
Finding
The trigger conditions are overly broad and mismatched to the skill's real scope, increasing the chance the agent invokes a system-control skill in inappropriate contexts. Because the skill has exec capability and includes OS-level commands, accidental invocation can lead to unintended command execution, privacy exposure, or destructive actions on the host.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill exposes screenshot and clipboard operations without a prominent privacy warning near the capability description. In an agent context, these features can access sensitive on-screen data and clipboard contents, creating a real risk of inadvertent collection or disclosure of secrets, personal data, or credentials.

Tool Parameter Abuse

High
Category
Tool Misuse
Content
| 复制到剪贴板 | `echo x \| pbcopy` |
| 粘贴 | `pbpaste` |
| 打开 Finder | `open .` |
| 清空废纸篓 | `rm -rf ~/.Trash/*`(先确认) |

## 安全操作规范
Confidence
98% confidence
Finding
rm -rf ~

Tool Parameter Abuse

High
Category
Tool Misuse
Content
| 复制到剪贴板 | `echo x \| pbcopy` |
| 粘贴 | `pbpaste` |
| 打开 Finder | `open .` |
| 清空废纸篓 | `rm -rf ~/.Trash/*`(先确认) |

## 安全操作规范
Confidence
98% confidence
Finding
rm -rf ~/.Trash/

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.