video-to-superprompt

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow ingests the reference video content at runtime—via locally uploaded files, extracted browser-visible videos, or user-supplied URLs—and those videos can be authored by outsiders, so their on-screen text/visual descriptions become free text in the agent’s LLM context through frame extraction/analysis.