Human Traits Learning
Security checks across malware telemetry and agentic risk
Overview
This skill is a documentation-only guide for AI behavior training, with no executable code or hidden system access found.
This appears safe to install as reference material. Users should still treat its personalization and user-pattern analysis guidance as opt-in: only allow an agent to analyze preferences or adapt behavior when that is clearly requested and transparent.
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.