04-script-video-global

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

• Third-party content exposure detected (high risk: 1.00). The skill's instructions explicitly tell the agent to ingest and "Analyze structure" and "Extract technique" when "the user shares a reference video link" (see the "Reverse-engineer viral videos (research only)" section of SKILL.md), which means the agent will fetch/read user-provided public video pages and use that untrusted third-party content to influence scripting decisions.