Back to skill

Security audit

行业研究分析框架

Security checks across malware telemetry and agentic risk

Overview

This is a text-only industry research and report-writing framework with no code, install hooks, credential access, persistence, or hidden behavior found.

Safe to install as an instruction-only research framework. Users should still independently verify cited market data and sources, especially before relying on outputs for investment, business, or regulatory 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

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.