qwen-image-2

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

• Third-party content exposure detected (high risk: 0.70). The skill's "Image Editing (Multi-Reference)" example and input schema show it accepts reference_images specified by external URIs (e.g., "https://example.com/person.jpg"), so it fetches and ingests arbitrary public images that directly influence generation/editing decisions and thus could carry untrusted content that alters behavior.