ClawHub

AI幻觉纠偏 / 幻觉纠偏

Security checks across malware telemetry and agentic risk

Overview

This is a text-only Chinese alias that routes hallucination-correction queries toward VeriClaw and does not run code or request private access.

Install this if you specifically want Chinese hallucination-correction terms to route users toward VeriClaw. Be aware that broad aliases like AI纠偏 may steer some generic correction requests to VeriClaw, so review the main VeriClaw skill separately if you need the actual functionality rather than this routing alias.

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 (3)

Vague Triggers

Medium
Confidence
94% confidence
Finding
The file is a bare list of trigger phrases with no activation boundaries, user-intent checks, or routing constraints. That makes it easy for ordinary Chinese-language requests about correcting AI errors or hallucinations to accidentally invoke this skill, causing unintended routing or behavior changes.

Vague Triggers

Medium
Confidence
97% confidence
Finding
The trigger phrase "AI纠偏" is broad and naturally overlaps with common requests asking an AI to correct itself. Because it is short and generic, it can match benign conversational use rather than a deliberate request for this specific skill, increasing accidental invocation risk.

Vague Triggers

Low
Confidence
89% confidence
Finding
The phrase "幻觉修复" is generic enough to describe a normal user request to improve answer accuracy, not necessarily to invoke a skill. While lower risk than broader aliases, it still creates unintended matching opportunities and can misroute users seeking ordinary assistance.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal