gpt-image-2

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

• Third-party content exposure detected (high risk: 0.90). The skill's edit endpoint explicitly accepts up to 10 user-supplied public image URLs (the "images" field in SKILL.md) which are fetched by the RunComfy model server and used as reference material—this clearly ingests untrusted third-party content that can influence model decisions and enable image-based prompt injection.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill's edit endpoint requires user-supplied, publicly fetchable image URLs (e.g., "https://...subject.jpg" / "https://...room.jpg") which are fetched at runtime by the RunComfy model server and can perform image-based prompt injection that directly influences model outputs.