vally-tests

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

• Third-party content exposure detected (medium risk: 0.30). In corpus-import mode, scripts/import_corpus.py reads outsider-supplied CSV/XLSX row text (especially the prompt field) via read_rows()→import_corpus()→safety_check(row["prompt"], ...), so the workflow ingests free text directly from the user-provided corpus.