Security audit
machine-learning-engineer
Security checks across malware telemetry and agentic risk
Overview
This is an instruction-only machine learning deployment advisor with no executable code or hidden behavior found.
Reasonable to install as an ML deployment advisor. Review and approve any suggested production infrastructure, registry, CI/CD, autoscaling, telemetry, or secret-management changes before running them, preferably in staging first.
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.