AI幻觉纠偏 / AI纠偏

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:35 PM UTC
• Version:
  0.1.10

Overview

This is an instruction-only Chinese alias that routes AI-correction and hallucination queries to VeriClaw, with no executable code or credential access.

Install this only if you want Chinese AI-correction or AI-hallucination terms to route toward VeriClaw. For the actual tool experience, prefer the main vericlaw skill/page; be aware that generic phrases may invoke this alias unexpectedly.

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: 86% confidence
• Finding: The trigger phrase `AI纠偏` and the surrounding alias language are broad enough to match normal user discussion about AI mistakes rather than an explicit request to invoke this specific skill. That can cause unintended routing to this alias or its canonical target, creating confused-deputy behavior and misdirecting users away from their original task.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The `Use it when the problem is described as` section enumerates many generic phrases like `AI纠错`, `AI校正`, and `AI幻觉`, but does not define when the alias must not activate. In an agent ecosystem, this broad matching can capture ordinary troubleshooting or conversational contexts and redirect users to external pages or installation instructions without clear consent.

VirusTotal

60/60 vendors flagged this skill as clean.