ClawHub

z-card-image

Security checks across malware telemetry and agentic risk

Overview

This appears to be a legitimate local image-generation skill, though it should be used carefully because it runs a browser and writes image files.

Install only if you are comfortable with a skill that runs local Python scripts and headless Chrome to render PNGs. Prefer workspace-relative output paths, avoid using sensitive local files as icons or avatars, and treat untrusted Markdown/HTML content cautiously because it is rendered in a browser.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (3)

Context-Inappropriate Capability

Medium
Confidence
83% confidence
Finding
The skill launches Chrome/Chromium in headless mode with the --no-sandbox flag to render attacker-controlled HTML content. Because the content is generated from untrusted text and markdown and then opened in a real browser process, disabling the browser sandbox increases the impact of any browser engine bug, local file access abuse, or active-content issue during rendering.

Context-Inappropriate Capability

Medium
Confidence
88% confidence
Finding
The skill executes external local binaries (Chrome/Chromium and optionally ffmpeg), which gives it capabilities beyond simple templating and file writing. In an agent-skill context, undocumented binary execution increases attack surface, may bypass sandbox expectations, and can expose local files or system resources if the environment is permissive.

Vague Triggers

Medium
Confidence
80% confidence
Finding
The routing rule says that any input resembling 'post-type data' and wanting X-style output should invoke `x-like-posts`, which is broad enough to capture many loosely related requests. Overly broad invocation criteria can cause the skill to process unexpected untrusted content and trigger code/file operations in contexts where a safer or non-executing path would have been more appropriate.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal