analyze-content

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

• Third-party content exposure detected (medium risk: 0.30). The skill takes a user-provided URL (or URLs pulled from the user’s Supabase “content” table) and then scrapes/reads the resulting captions, transcripts, slide text, and email-like page text for Claude processing at runtime (e.g., carousel slide text via Apify and longform captions via yt-dlp), meaning outsider-authored free text can be ingested directly without selecting a specific item first.