vally-tests

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

• Third-party content exposure detected (high risk: 0.85). In corpus-import mode, the skill reads outsider-authored free text from a caller-supplied CSV/XLSX corpus (scripts/import_corpus.py: read_csv_rows/read_xlsx_rows), then embeds that prompt text into emitted YAML stimulus blocks (build_patch_entry), which becomes LLM-readable context during later Vally eval runs.