ClawHub

food-safety-legal

Security checks across malware telemetry and agentic risk

Overview

This is a documentation-only Chinese food-safety compliance helper with no hidden code or data-sharing behavior found.

Install is reasonable if you want help reviewing Chinese food-safety legal compliance. Share only the product and business materials needed for the review, confirm the target jurisdiction, and verify important legal conclusions against current regulations or a qualified legal professional.

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 (2)

Vague Triggers

Medium
Confidence
86% confidence
Finding
The skill’s trigger description is broad enough to overlap with common legal/compliance questions, which can cause unintended invocation of this specialist skill outside clearly scoped food-safety contexts. In an agent setting, misrouting can lead to overconfident domain-specific guidance, incomplete jurisdiction checks, or collection of unnecessary business-sensitive documents from users who did not intend to engage this workflow.

Natural-Language Policy Violations

Medium
Confidence
79% confidence
Finding
The skill is hard-coded to Chinese language and PRC regulatory analysis without offering a language or jurisdiction check, which risks giving users incorrect compliance advice when their product, market, or question involves another country. In a legal-compliance skill, jurisdictional mismatch is particularly risky because users may rely on output for labeling, advertising, or import/export decisions that differ materially across regions.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal