Ingest a large or binary local document before analysis: PDF, slide deck, spreadsheet, CSV dump, or long document. Convert it to indexed lightweight Markdown/CSV artifacts. Use when an analysis request touches one; don't use when a current verified ingestion already exists.
