flux-image

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

• Third-party content exposure detected (high risk: 0.70). The SKILL.md explicitly shows "Image-to-Image" and "With LoRA Custom Styles" examples that accept arbitrary public URLs (image_url, lora_url) which the service will fetch and ingest from third-party sites, allowing untrusted user-provided content to influence model behavior and generation outputs.