ClawHub
Back to skill

Security audit

Card Image Builder

Security checks across malware telemetry and agentic risk

Overview

This image-rendering skill is mostly purpose-aligned, but it needs review because its network use, command execution, and bulk file-writing behavior are not clearly scoped.

Install only if you are comfortable letting the agent run local Python/rendering commands and create batches of image/template files. Use explicit input and output directories, avoid broad recursive inputs, review remote image URLs before rendering, and do not rely on the claim that the workflow never touches the network.

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 (4)

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The skill claims it is a local-only renderer that does not involve network requests, yet elsewhere documents network diagnostics, external downloads, remote avatar URLs, and LLM/API dependencies. This kind of contradictory security description can cause users or agents to make unsafe trust decisions, such as allowing execution under assumptions of no network exposure when the workflow may in fact touch remote resources.

Context-Inappropriate Capability

Medium
Confidence
84% confidence
Finding
The skill is presented as an image-generation tool but declares EXEC capability and includes guidance that can drive arbitrary shell/Python command execution beyond narrowly scoped rendering. When an agent is allowed to execute commands from a loosely constrained skill, the attack surface expands from content rendering to system-level actions, increasing the chance of misuse or prompt-to-command abuse.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill describes generating output images but does not adequately warn that it will write files to disk, including batch outputs and temporary artifacts. In agent contexts, undisclosed filesystem modification is risky because users may authorize the skill without understanding that it can create or overwrite files in working directories.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The batch-generation feature explicitly mentions directory-recursive processing but does not warn about large-scale file creation, resource consumption, or unintended traversal of broad input trees. In an agent setting, recursive processing can magnify mistakes into widespread disk usage, accidental processing of sensitive content, or operational disruption.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.