ai-video-generation

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly shows runtime ingestion of arbitrary external URLs (e.g., Image-to-Video falai/wan-2-5 with "image_url", Fabric/OmniHuman/PixVerse examples with "image_url" and "audio_url", and media-merger with "videos" URLs), which means the agent will fetch and process untrusted public third‑party content that can materially influence model outputs and downstream actions, enabling indirect prompt injection.