Back to skill

Security audit

Retrospective Analysis

Security checks across malware telemetry and agentic risk

Overview

This skill is a straightforward local retrospective-analysis helper with explicit JSON save/load behavior and no evidence of hidden network, credential, or background activity.

Installers should understand that retrospective content may include internal project/process details, and exported JSON reports will be written wherever the caller specifies. Avoid saving sensitive retrospectives to shared or unintended paths.

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

57/57 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.