jimeng

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

• Third-party content exposure detected (high risk: 0.70). The SKILL.md explicitly requires an "image" parameter to be a publicly accessible URL for image-to-image generation (Image-to-Image section), meaning the agent will fetch and ingest arbitrary public images (untrusted third-party content) that directly influence model outputs and subsequent behavior.