Social Alignment
Security checks across malware telemetry and agentic risk
Overview
This is a disclosed alignment-check skill with optional local decision memory and no evidence of hidden data theft or destructive behavior.
Install this only if you want the agent to pause and reason through significant actions before proceeding. Verify the external social-alignment Python package before running it, use an isolated environment for testing, and enable persistent decision memory only in a file location you control because it may reveal patterns about your decisions.
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
64/64 vendors flagged this skill as clean.