tutorial

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

• Third-party content exposure detected (high risk: 1.00). The tutorial explicitly fetches and processes public, user-generated corpus content (e.g., scry.mv_lesswrong_posts, scry.mv_eaforum_posts, scry.mv_hackernews_posts and other scry.entities via /v1/scry/query in Steps 3–5), and instructs the agent to read/extract/rerank those excerpts (Step 5) and use them to create shares/judgements, so untrusted third‑party content can materially influence LLM-driven ranking and subsequent actions.