PDF extract
- Confirm the file path and the desired output (text, tables, filled form, merge).
- Prefer
scripts/extract.pyover ad-hoc one-liners. - For form field maps see FORMS.md.
- Return the artifact path and a one-line summary of pages processed.
Do not load every page into context. Stream through the script and keep only the rows you need.
python scripts/extract.py input.pdf --tables
Safety
Never upload the PDF to a third-party API. Confirm page count before writing.