brand-protection
Security checks across malware telemetry and agentic risk
Overview
This is a markdown-only brand-protection guide for collecting evidence and filing takedown reports, with no hidden code, persistence, credential use, or destructive behavior found.
Before installing, be aware the skill may use local project-context files to identify your brand and official domain. Review those files for sensitive details, redact private customer/payment data before submitting reports, and verify URLs carefully before filing abuse, takedown, or legal complaints.
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
65/65 vendors flagged this skill as clean.