Ai Augmentation Not Automation
Security checks across malware telemetry and agentic risk
Overview
This is a text-only leadership guidance skill with no code, install hooks, hidden data access, or privileged behavior.
Safe to install as an advisory business strategy skill; verify any cited business examples or statistics before relying on them, and avoid placing sensitive company details into prompts unless your model/runtime is approved for that data.
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.