ClawHub
Back to skill

Security audit

Deep Dialogue

Security checks across malware telemetry and agentic risk

Overview

This is a text-only reflective dialogue skill with no code or system access, but it should not be treated as therapy or crisis support.

Install only if you want an agent to engage in deeper reflective dialogue. Do not use it as a substitute for licensed mental health care, diagnosis, emergency help, or crisis support. Avoid using it for self-harm risk, abuse, psychosis, severe depression, trauma crisis, or acute distress; use professional or emergency resources instead. Verify the GitHub source before installing because the README clone command is not pinned to a reviewed commit.

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 (4)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The README says the skill triggers on broad 'philosophical/psychological conversations,' which can cause it to activate in many sensitive contexts without clear boundaries. In a skill that performs emotionally probing, dialectical challenge, and quasi-therapeutic guidance, ambiguous invocation increases the chance of unscoped use during crisis, trauma, or other high-risk situations where this style may be inappropriate.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill presents psychologically intensive techniques drawn from CBT, ACT, existential challenge, and Jungian interpretation, but provides no warning about limits, contraindications, or the need for caution in vulnerable users. Because the examples include emotionally forceful reframing and challenge, users may receive pseudo-therapeutic guidance in situations involving depression, trauma, suicidality, or coercive dynamics, increasing the risk of harm or delayed access to appropriate help.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill explicitly targets personal psychological struggles, difficult emotions, and existential distress, but it provides no boundary-setting that it is not a substitute for mental-health care, crisis support, or licensed professional advice. In this context, users may over-rely on the skill during vulnerable moments, leading to delayed escalation, inappropriate guidance, or unsafe handling of crisis-adjacent situations.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The activation description is broad enough to trigger on many ordinary conversations about feelings, beliefs, or meaning, without clear boundaries for when deep psychological probing is appropriate. This increases the chance of the agent entering a quasi-therapeutic mode unexpectedly, gathering sensitive information or applying challenging frameworks in situations that would be better handled by normal conversational support or safer triage.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.