ai-video-gen

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

• Third-party content exposure detected (high risk: 0.70). The SKILL.md explicitly accepts arbitrary reference_image_url and tail_image_url (e.g., in the "Request Fields" and "Image-to-Video" examples) which the HeyGen workflow will fetch/ingest from public URLs and use to guide video generation, so untrusted third‑party content can materially influence the agent's outputs.