make-a-video

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly asks for and instructs the agent to fetch and "watch / read" user-supplied reference video URLs (see references/style-intake.md: "If URLs, fetch them" and Gate 3's "Reference videos for vibe? (URLs/paths · none)"), and then extract palette, pacing, and motion vocabulary from those arbitrary external resources to drive build decisions, so untrusted third‑party content can materially influence tool use and next actions.