ClawHub
Back to skill

Security audit

Anti-Hallucination 防胡编卫士

Security checks across malware telemetry and agentic risk

Overview

This prompt-only anti-hallucination skill is purpose-aligned and non-executable, though it may make an assistant more cautious or trigger verification more often than expected.

Install this if you want stricter source-backed answers. Expect more refusals or uncertainty when records are missing, and review the always-on, strict mode, trigger phrases, and optional deep-check behavior so the assistant does not search files or the web more often than you intend.

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 (1)

Vague Triggers

Medium
Confidence
90% confidence
Finding
The trigger list contains very common conversational phrases such as 'really?', 'are you sure', and broad Chinese equivalents. In systems that auto-activate skills based on phrase matching, this can cause unintended invocation, altering agent behavior unexpectedly and creating prompt-routing instability or denial-of-service-style interference with normal interactions.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.