ClawHub
Back to skill

Security audit

Alibaba Cloud AI Video LivePortrait

Security checks across malware telemetry and agentic risk

Overview

This skill is a straightforward Alibaba Cloud LivePortrait helper, with disclosed but privacy-sensitive handling of portrait and audio URLs.

Install this only if you intend to process portrait and voice media with Alibaba Cloud Model Studio. Use media you have rights and consent to use, prefer short-lived or controlled URLs, avoid embedding secrets in URL query strings, use a least-privileged DashScope credential, and delete output/aliyun-liveportrait artifacts when they contain private links.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Lp3

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding
The skill documentation describes capabilities that write files under an output directory and rely on network access to public URLs and Alibaba Cloud APIs, but it does not declare any permissions or clearly scope those capabilities. This creates a trust and review gap: users may invoke a skill with filesystem and outbound network behavior they were not explicitly warned about, increasing the chance of unintended data handling or exfiltration through normal operation.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill explicitly requires public HTTP/HTTPS image and audio URLs and says to record the exact portrait/audio URLs, but provides no privacy or data-sensitivity warning. This can expose personal portraits, voice samples, and signed or guessable resource links to third parties and to local output artifacts, especially because talking-head generation inherently processes biometric and voice-related content.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.