z-video-study-webpage-qwen
Pass
Audited by Gen Agent Trust Hub on Jun 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill interacts with Aliyun DashScope (dashscope.aliyuncs.com), a well-known AI service, to perform multimodal analysis of video content.
- [COMMAND_EXECUTION]: The skill uses subprocess calls to execute ffmpeg and ffprobe for extracting video frames and metadata. This is standard behavior for the skill's stated purpose of video study.
- [CREDENTIALS_SAFE]: The implementation follows security best practices by reading the API key from environment variables, scrubbing secrets from error messages using regex patterns, and performing a final scan of all output files to ensure no credentials have been leaked.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from video transcripts which could contain malicious instructions. However, it mitigates risk by using structured JSON schemas for model responses and applying HTML escaping when rendering the final study webpage.
Audit Metadata