image-edit

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

• Third-party content exposure detected (high risk: 0.80). The skill requires publicly-fetchable HTTPS image/mask URLs (see the schema fields like "image_urls"/"images"/"mask_image") that the RunComfy model server fetches and uses to drive edits (Security & Privacy notes the server fetches external URLs and calls out "image-based prompt injection" as a known risk), so it clearly ingests untrusted third‑party content that can influence model behavior.