needscope

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

• Third-party content exposure detected (high risk: 0.85). Step 4/5/6/7 ingest and LLM-processes the user-provided consumer text samples (e.g., coded_items[].text and evidence quotes) which are outsider-authored free-form comments; this text is then included in the agent’s LLM context during semantic coding and report generation.