train

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

• Third-party content exposure detected (high risk: 0.70). Outsider free text is ingested via runtime file I/O: scripts/analyze_results.py reads results.tsv using csv.DictReader and prints fields like description/experiment_id into stdout (LLM context), and results.tsv is not shown as user-authored/bundled reference.