PDF Extract
Steps
- Run
scripts/extract.py <file.pdf>to get raw text and tables as Markdown. - Review headings; PDFs often lose structure — restore
##sections from font-size hints in the output. - Return the Markdown to the user. For very long documents, summarize per section first and ask before dumping everything.
Notes
- Scanned PDFs have no text layer; tell the user OCR is required rather than returning empty output.
- Never send the document contents to any external service.