ClawHub
Back to skill

Security audit

structured-output

Security checks across malware telemetry and agentic risk

Overview

This is a lightweight reporting-format skill with no executable files or persistence, though it includes a few unnecessary generic command-verification prompts.

Before installing, be aware that this skill may activate for ordinary formatting or reporting requests. Treat the generic '--help' verification lines as optional only when a specific command is already relevant to your task; the skill does not require installing or running anything by itself.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
This formatting-only skill instructs the agent to run a command for verification, which is unnecessary to its stated purpose and can induce unintended tool execution. Even a seemingly harmless '--help' invocation expands the skill's operational scope from presentation guidance into command execution, creating avoidable risk and confusion during security-sensitive workflows.

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The action-assignment section repeats an instruction to run a command despite the skill being documentation-oriented, encouraging unnecessary execution behavior. In an agent environment, normalizing such instructions can erode the boundary between text formatting tasks and tool use, increasing the chance of misuse or execution in the wrong context.

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The appendix step again directs command verification that does not align with the skill's declared behavior, making the instruction contradictory and potentially unsafe. Repeated embedded execution cues are especially problematic in reusable skills because they can condition agents or operators to perform commands without clear necessity.

Vague Triggers

Medium
Confidence
85% confidence
Finding
The trigger keywords and auto-load conditions are broad enough to match many ordinary requests involving formatting or reporting, causing the skill to activate in situations where it may not be needed. Overbroad activation increases attack surface by making it easier for adversarial or unrelated prompts to pull this skill into workflows, including contexts where its embedded execution guidance may then be surfaced.

VirusTotal

59/59 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.