clipsketch-ai-video-storyboard

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly imports videos from public platforms (Bilibili and Xiaohongshu per the "Video Import" URL pattern section) and passes extracted frames into analyzeFrames/generateStoryboard (SKILL.md core workflow) so untrusted, user-generated content is ingested and can materially influence model prompts and downstream actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly accepts and resolves external video URLs (e.g., https://b23.tv/ or https://bilibili.com/video/...), captures frames from those remote videos at runtime, and injects the resulting image data directly into model.generateContent calls, so remote content fetched from those URLs directly controls the AI prompts/inputs.