Return the extraction result enriched with per-leaf source provenance.
Each extracted leaf value is wrapped as where source contains citation content, surrounding context, and a format-specific anchor (bbox for PDFs, cell ref for spreadsheets, text span for plain text, etc.).
Successful Response
ID of the extraction
Detected document type of the source file
pdf, image, csv, xlsx, docx, txt File metadata (id, filename, mime_type). Use the Files API to download the content.
Original extraction output
Same shape as extraction but leaves are {value, source} objects
"extraction.sources"