ClawHub
Back to skill

Security audit

Identity Primitive Chain

Security checks across malware telemetry and agentic risk

Overview

This appears to be a prompt-only meta-skill with transparency ambiguity, but no evidence of malware, data theft, persistence, or destructive behavior.

Install only if you want a meta-skill that may influence how your agent decomposes and routes tasks. Ask the agent to show concise execution summaries when using it, and avoid relying on it for sensitive compliance, security, or high-impact decisions unless you can review the applied steps.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Intent-Code Divergence

Medium
Confidence
89% confidence
Finding
The skill defines conflicting transparency rules: it says input/processing/output details are hidden by default, but elsewhere requires the complexity assessment derivation to be shown to the user. Ambiguous disclosure behavior can cause the agent to inconsistently reveal internal reasoning or withhold required justification, which is risky in a meta-skill that orchestrates other behaviors.

Intent-Code Divergence

Medium
Confidence
86% confidence
Finding
The read-only mode forbids interaction (I), but the skill also claims users can request execution details. This contradiction can produce unsafe or undefined behavior in constrained modes, where the agent may either ignore a safety restriction or improvise a policy bypass to satisfy the request.

Vague Triggers

Medium
Confidence
80% confidence
Finding
The trigger phrases are broad and loosely bounded, increasing the chance the skill activates in contexts where it was not intended. Because this is a meta-skill for task decomposition and identity layering, accidental activation could override more specific instructions, alter execution flow, or introduce hidden orchestration logic into unrelated tasks.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The skill explicitly states that execution input/processing/output details are hidden by default unless the user asks to see them. In an agent skill, default opacity reduces oversight, makes it harder to audit unsafe routing or transformations, and can conceal harmful behavior or policy-violating intermediate steps from users and reviewers.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.