Back to skill

Security audit

performance-optimization

Security checks across malware telemetry and agentic risk

Overview

This skill provides performance-optimization guidance and examples without hidden execution, persistence, or unrelated data access.

Installers should expect this skill to guide an agent through profiling and performance fixes, including optional CI performance tooling. Review any proposed profiling commands or code changes before running them in production, but the artifact itself does not show hidden or unsafe behavior.

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.

View on VirusTotal

Static analysis

No suspicious patterns detected.