magic-data-synthesis

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

• Third-party content exposure detected (medium risk: 0.65). The required runtime path for this skill is the LLM prompt construction in scripts/batch_synthesize.py (build_prompt_for_row()), which injects per-row free-text from the input dataset (seed columns and other context columns) into the LLM prompt; if that dataset/columns originate from outsiders, this becomes an indirect prompt-injection channel.