Qwen Video (Wan)

• Outcome: Review
• Latest audit: May 28, 2026, 11:31 AM
• Version:
  1.1.0

Overview

This video-generation skill appears purpose-aligned, but it disables HTTPS certificate checks while using an API key and has weak request/output safety controls.

Install only if you are comfortable patching or reviewing the scripts first. Remove curl -k, use a limited DashScope API key if possible, avoid sensitive prompts or audio URLs, and choose an output path where overwriting an MP4 would not matter.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Context-Inappropriate Capability

Medium

• Confidence: 99% confidence
• Finding: The script disables TLS certificate verification with curl -k while sending an Authorization bearer token and user-supplied content to a remote API. This enables man-in-the-middle interception or tampering, which can expose the API key, alter requests, or forge responses; the video-generation context does not justify weakening transport security.

Missing User Warnings

Low

• Confidence: 88% confidence
• Finding: The description says the skill will download the MP4 locally but does not clearly warn that it writes to an arbitrary user-specified local path, including a Windows Desktop path from WSL2. While this is not code execution, it can cause unexpected local filesystem modification and weakens informed consent for potentially sensitive destinations.

Missing User Warnings

Medium

• Confidence: 98% confidence
• Finding: The script downloads remote content with curl -k, which disables TLS certificate verification and allows a man-in-the-middle attacker to substitute arbitrary content for the video download. Because the file is then written directly to a user-specified path, this can silently overwrite local files with attacker-controlled data, which is especially risky in an agent skill that automatically retrieves and stores outputs.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: Certificate verification is explicitly disabled for an outbound HTTPS request without any warning or opt-in, so users may assume the API call is protected when it is not. Because the request includes a bearer token, an attacker on the network path could impersonate the API endpoint and capture credentials or manipulate job submission/results.

VirusTotal

64/64 vendors flagged this skill as clean.