Zhu Xiaohu Perspective V2

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only persona skill for investment/business analysis, with no code execution, file access, credentials, or data transfer behavior.

Install this only if you want an opinionated role-play lens for investment and business analysis. Treat outputs as perspective and not financial advice, and use the documented exit phrases if the persona activates unexpectedly.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
Findings (2)

Vague Triggers

Medium
Confidence
91% confidence
Finding
The trigger list includes broad phrases such as common discussion terms around investing and perspective-taking, which can cause unintentional activation of the skill when a user is merely discussing the subject. This is dangerous because it can unexpectedly force persona behavior and alter response style or framing without clear user consent, reducing reliability and user control.

Natural-Language Policy Violations

Medium
Confidence
94% confidence
Finding
The skill requires the assistant to immediately respond in a fixed persona and style, including first-person roleplay and specific rhetorical patterns, without confirming the user's intent or offering an alternative. This is risky because it can override normal assistant behavior, create misleading anthropomorphic framing, and make the model less responsive to user preferences or safety-sensitive context.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal