Back to skill

Security audit

Design Assets

Security checks across malware telemetry and agentic risk

Overview

This is a design helper with local image-processing examples that match its stated purpose and do not show hidden or harmful behavior.

Safe to install as a design workflow reference. Run batch image commands only in the intended directory or on copies, and avoid sending confidential design prompts or brand material to any AI image provider unless that provider is approved for your use.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.