genmedia-ref

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

• Third-party content exposure detected (high risk: 0.70). The SKILL.md explicitly states the CLI "walks the result payload and downloads any object with a url: string starting with http(s)://" and instructs agents to use --download to fetch those external media URLs, meaning the agent will fetch and consume arbitrary third-party HTTP(S) content from model outputs (untrusted/user-provided) as part of its workflow.