ClawHub

Stock Images

Security checks across malware telemetry and agentic risk

Overview

This is a markdown-only stock-image reference skill with some transparency notes around third-party image URLs, but no evidence of hidden execution, credential theft, destructive behavior, or persistence without user direction.

Reasonable to install for mockups and placeholder images. Avoid using sensitive search terms or private image URLs with third-party services, verify licensing before production use, do not paste real API keys into shared chats or files, and only enable the optional memory file if you want preferences saved locally.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Vague Triggers

Medium
Confidence
93% confidence
Finding
The instruction to "read this silently when user first asks for stock images" creates a vague activation boundary that can cause the skill to engage on loosely related prompts. Over-broad triggering can lead to unintended behavior, hidden context injection, or use of this skill when the user did not explicitly request it, which reduces transparency and control.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill explicitly directs hidden setup and internal retention of user preferences without informing the user, which creates a transparency and privacy issue. Even if the retained data is low sensitivity, silent persistence of behavioral preferences can violate user expectations and enable unnoticed profiling across interactions.

Missing User Warnings

Low
Confidence
88% confidence
Finding
The file encourages direct use of Unsplash source URLs but does not warn that each request contacts a third-party service, exposing client IP address, user agent, referrer, and timing metadata to that external provider. It also omits that returned images may change over time due to redirect-based dynamic selection, which can create privacy, reliability, and content-governance issues in downstream applications.

External Transmission

Medium
Category
Data Exfiltration
Content
```bash
# ReSmush.it (no auth needed)
curl -X POST "http://api.resmush.it/ws.php?img=https://example.com/image.jpg"
```
Confidence
82% confidence
Finding
http://api.resmush.it/

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal