First Principles Thinking
Security checks across malware telemetry and agentic risk
Overview
This appears to be a documentation-only thinking-method skill with no code execution, data access, network use, or persistence.
This skill is suitable to install if you want a reusable reasoning framework. As with any skill, review the visible instructions, but there is no evidence here of code execution, sensitive data access, network transfer, or persistence.
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
63/63 vendors flagged this skill as clean.