Su Hua Perspective V2

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:36 PM UTC
• Version:
  1.0.0

Overview

This is a prompt-only Su Hua-inspired consulting persona with no code or system access, but it can blur attribution by speaking in first person.

Install only if you want a Su Hua-inspired strategy persona. For safer use, trigger it explicitly and keep a visible disclaimer in responses so generated advice is not mistaken for Su Hua’s actual views or endorsement.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

• Confidence: 96% confidence
• Finding: The trigger rules are explicitly broadened to activate on generic topical queries like algorithms, lower-tier markets, community ecology, and trust-based e-commerce, even when the user does not request this persona. That creates unintended persona takeover risk, where normal queries can be answered through a roleplay lens that injects bias, reduces transparency, and may override more appropriate assistant behavior.

Natural-Language Policy Violations

High

• Confidence: 99% confidence
• Finding: The skill directs the agent to answer in first person as if it were Su Hua and to avoid meta-analysis, which increases the risk of misleading users about identity and source authenticity. Although there is a one-time disclaimer elsewhere, persistent first-person impersonation can still cause users to attribute statements, advice, or factual claims to the real person, especially in business or strategic contexts.

VirusTotal

62/62 vendors flagged this skill as clean.