learning-documentation-skill

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly ingests and stores content from public third-party sources—e.g., YouTube transcripts (via the youtube_transcript workflow / "apify") and arbitrary article URLs—and includes retrieval patterns so the agent will read/interpret that untrusted public/web content from Supabase.