ClawHub
Back to skill

Security audit

Qwen视频智能分析

Security checks across malware telemetry and agentic risk

Overview

This video-analysis skill is purpose-aligned overall, but it needs Review because it handles local media and API keys with unclear third-party data disclosure and inconsistent credential instructions.

Before installing, treat this as a third-party cloud video-analysis skill: only use it with media you are allowed to send to Alibaba Cloud DashScope/Qwen, avoid sensitive or regulated videos unless you have reviewed the provider's data terms, and fix the API-key instructions to one protected config path before use.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The skill states the API key is read from `$HOME/.skill-platform/skill-platform.json`, but the authentication section later instructs users to read and populate `$HOME/.json`. This inconsistency can cause users to place secrets in the wrong file, increasing the chance of accidental exposure, misconfiguration, or use of less-protected ad hoc secret storage.

Intent-Code Divergence

Medium
Confidence
92% confidence
Finding
The security section recommends environment variables for API keys, while the rest of the skill says the key is only read from config files. Conflicting secret-handling guidance leads to operator confusion, insecure workarounds, and failed expectations about where sensitive credentials are stored and protected.

Vague Triggers

Medium
Confidence
81% confidence
Finding
The activation language is overly broad and ambiguous, describing use cases like video processing, audio editing, media conversion, and dubbing generation even though the skill is for video analysis. Overbroad triggers can cause the agent to invoke this skill in unintended contexts, increasing the chance of sending unrelated local files or URLs to external processing.

Missing User Warnings

High
Confidence
97% confidence
Finding
The documentation does not clearly warn that local video content and remote video URLs are transmitted to a third-party API for analysis. This is dangerous because users may provide sensitive or regulated media without informed consent, creating privacy, confidentiality, and compliance risks when data leaves the local environment.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.