ClawHub
Back to skill

Security audit

视频

Security checks across malware telemetry and agentic risk

Overview

This skill is for video generation, but it gives agents broad local command execution through a custom TTS command without clear safety boundaries.

Review this before installing. Use it only in an environment where local command execution is contained, verify the source of any generate.py implementation, and do not let untrusted Markdown text or user input populate a custom TTS command template. Limit API keys and file access to what the video job actually needs.

SkillSpector

Vulnerability Patterns
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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)

Intent-Code Divergence

Medium
Confidence
90% confidence
Finding
The skill claims command execution occurs in a safe sandbox, yet the same document exposes exec capability and a user-supplied custom TTS command template without any concrete sandboxing boundaries, restrictions, or enforcement details. This can mislead users and downstream agents into trusting execution paths that may actually run arbitrary local commands with host privileges.

Missing User Warnings

High
Confidence
96% confidence
Finding
The documented --tts-command feature allows a caller to supply a command template containing user-controlled placeholders, which can enable arbitrary command execution if passed to a shell or unsafe subprocess invocation. Omitting an explicit warning and safe-construction guidance is dangerous because this skill operates in an agent context with read/write/exec tools, making misuse materially more likely and more damaging.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.