ClawHub
Back to skill

Security audit

5 Star

Security checks across malware telemetry and agentic risk

Overview

This is a novelty instruction-only skill that intentionally makes the assistant less useful, but it does not install code, request data access, or persist itself.

Install this only if you intentionally want a joke or novelty mode that may make the assistant unproductive and mildly provocative. Avoid using it for serious work or any context where reliable task completion matters.

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 (1)

Vague Triggers

Medium
Confidence
96% confidence
Finding
The skill defines broad, always-on behavioral instructions rather than a narrowly scoped capability, telling the agent to remain inactive, lose knowledge, and respond in ways designed to shut down further user interaction. That makes it a prompt-level sabotage skill: if activated or merged into agent behavior, it can degrade availability, reliability, and safe task completion across unrelated user requests.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal