ClawHub
Back to skill

Security audit

Ai Artist Workstation Pro

Security checks across malware telemetry and agentic risk

Overview

This skill is an AI image-order workflow, but it asks for broad agent capabilities and handles customer images through local storage and cloud delivery without enough user-control detail.

Review before installing. Use this only for image-generation order workflows, and avoid giving it unrelated agent-orchestration tasks. Before processing customer selfies or order data, require explicit consent for local storage, third-party image APIs, cloud-drive upload, and e-commerce message delivery, and define where files and retry queues are stored and when they are deleted.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The skill claims command execution is limited to a whitelist, but the document only declares broad exec capability and provides no concrete whitelist, validation rules, or enforcement mechanism. In an agent setting, this can create a false sense of safety and allow unsafe command construction or arbitrary command execution if later implementation follows the spec loosely.

Vague Triggers

Medium
Confidence
85% confidence
Finding
The activation text says to use the skill whenever AI model calling, dialogue, agent orchestration, or LLM application needs arise, which is far broader than the actual narrow image-generation workflow. Overbroad triggers can cause the agent to invoke a skill with read/write/exec capabilities in unrelated contexts, unnecessarily expanding the attack surface and risking misuse of local files, commands, or external APIs.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The workflow explicitly downloads generated images to local storage and delivers outputs through cloud-drive links, but it does not present a clear user-facing consent notice or data-handling policy at the point of use. Because inputs may include customer selfies and order metadata, silent local storage and third-party upload can expose sensitive personal data or create privacy/compliance issues.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.