videoagent-video-studio

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly accepts and forwards arbitrary public image/video URLs as inputs (see SKILL.md "image-to-video" examples and tools/generate.js requiring --image-url, plus proxy/models.js/refInput which sends image_urls/video_url to external model endpoints), so it ingests untrusted third-party content that can directly influence generation behavior.