ClawHub
Back to skill

Security audit

pet-food-compare

Security checks across malware telemetry and agentic risk

Overview

This appears to be a pet-food comparison guidance skill with health-adjacent advice and review analysis, but no evidence of hidden execution, credential use, persistence, or data exfiltration.

Install only if you want an informational pet-food comparison helper. Use it to organize product facts and reviews, but verify claims against labels, official product pages, and a veterinarian for allergies, obesity, urinary issues, prescription diets, or any medical condition.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Description-Behavior Mismatch

Medium
Confidence
86% confidence
Finding
The skill claims to remain neutral, but the '配料特征-动物情况匹配' section crosses into individualized or quasi-advisory guidance by mapping nutritional traits to animal conditions. In a pet-health context, this can steer purchase or feeding decisions without veterinary validation, especially for sensitive cases like allergies, obesity, urinary issues, or prescription diets.

Description-Behavior Mismatch

Medium
Confidence
78% confidence
Finding
The skill substantially expands scope from factual product comparison into review scraping and sentiment mining across e-commerce, social platforms, and communities. This increases exposure to unreliable, promotional, manipulated, or privacy-sensitive content, and can bias outputs away from verifiable facts despite the skill's stated neutrality.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.