image-generation

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly instructs uploading/hosting and passing arbitrary image URLs (e.g., "character_reference_image_url", "style_reference_image_url" in the Character Consistency Workflow and "frame0_url" for Luma) so the agent will fetch and use untrusted third-party images as inputs that materially influence generation behavior.