queryable-markdown

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/mdq.py, the required runtime workflow find/run/scan (collection search) ingests outsider-authored free text when it is part of user-provided Markdown inputs that are searched/matched without the agent first selecting a specific pre-known record (e.g., line_local_search_records scans all lines for a literal substring and returns line-local record contexts).