image-outpainting

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly ingests arbitrary external image URLs and style-reference images as part of its core workflow (see the "image_urls" fields in the Invoke examples and the "Security & Privacy" section noting "source image URLs and any style-reference images are untrusted"), so untrusted third‑party content is fetched and can materially influence model routing and editing behavior.