ClawHub
Back to skill

Security audit

Paid Measurement Loop

Security checks across malware telemetry and agentic risk

Overview

The skill is mostly coherent for paid-ad readbacks, but it directs automatic local ledger writes of business performance data without clear user confirmation or scoped write controls.

Review before installing. The skill does not show exfiltration or destructive behavior, but users should only run it where local ledger writes are acceptable and should confirm exactly what campaign data will be stored.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (2)

Description-Behavior Mismatch

Medium
Confidence
89% confidence
Finding
The skill’s stated role is analytical readback, but it instructs the agent to persist campaign data into a shared ledger via a command-line tool. That expands scope from read-only analysis to state mutation, which can create integrity, privacy, and cross-task contamination risks if untrusted inputs such as campaign names or exported values are recorded without strict validation.

Context-Inappropriate Capability

Medium
Confidence
93% confidence
Finding
The skill provides direct shell execution guidance using `python3 ... ledger.py` even though its main function is to compare paid-media performance and delegate ROI math. Embedding shell commands in skill instructions increases the chance that an agent will execute local tooling against attacker-controlled or malformed inputs, leading to unauthorized writes, path misuse, or unsafe command composition in downstream implementations.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.