ClawHub
Back to skill

Security audit

Cognitive Reframe

Security checks for vulnerabilities and agentic risk

Overview

This is a text-only CBT-style self-help skill with clear limits and no technical access, though users outside China should not rely on its hotline examples.

Install only as a self-help thought-reframing aid, not as mental health care. If a conversation involves self-harm, suicidal thoughts, severe or persistent symptoms, trauma, eating-disorder concerns, or immediate danger, use local emergency or crisis services instead of relying on this skill; users outside China should find crisis resources for their own region.

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
90% confidence
Finding
The trigger list includes very broad phrases like 'I feel terrible' and 'Things will never get better,' which can match ordinary distress or potentially serious mental-health disclosures without enough specificity. In a wellness skill, this increases the chance of unintended invocation and inappropriate routing into lightweight CBT reframing when a different response, escalation path, or no skill activation would be safer.

Natural-Language Policy Violations

Medium
Confidence
85% confidence
Finding
The examples and crisis resources are China-specific, but the skill does not clearly state that it is intended for users in China or offer region selection. In a mental-health-adjacent context, this can misdirect users to unavailable resources or culturally mismatched guidance during high-risk moments, reducing the effectiveness of safety escalation.

Static analysis

No suspicious patterns detected.