ClawHub
Back to skill

Security audit

Audience Mapper

Security checks across malware telemetry and agentic risk

Overview

This is a marketing research skill with a small scope inconsistency around named creator recommendations, but no hidden code, credential use, automatic execution, or unsafe persistence.

Safe to install for audience and niche research. Before using it, keep named creator outputs limited to public key-voice context or hand them off to a dedicated influencer-discovery workflow for actual shortlisting, vetting, and outreach decisions.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The skill metadata says this skill is not for finding specific creators to contract, but the niche-mode template explicitly asks for named creator recommendations and priority lists. That contradiction can cause downstream agents to perform disallowed influencer discovery tasks, expanding data collection and outreach targeting beyond the declared scope.

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The worked example reinforces the contradictory behavior by presenting specific creator identification as a normal expected output, despite the skill description prohibiting that function. Examples are high-authority guidance for agents, so this increases the chance the skill will be used to generate targeted creator lists and bypass intended task separation.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.