PDF To Md

PDF to Markdown

Convert scanned newspaper or magazine PDFs (FT, The Economist, etc.) to readable Markdown files using Tesseract OCR + Claude cleanup.

Trigger

Use when the user says:

• "convert this PDF to markdown"
• "extract text from this newspaper PDF"
• "I have a new FT / Economist PDF"
• "read this PDF as a note"

Usage

~/.venvs/marker/bin/python ~/.claude/skills/pdf-to-md/scripts/pdf_to_md.py \
  /path/to/newspaper.pdf \
  /path/to/output.md

Skip the Claude cleanup step (raw OCR only, faster):

~/.venvs/marker/bin/python ~/.claude/skills/pdf-to-md/scripts/pdf_to_md.py \
  /path/to/newspaper.pdf \
  /path/to/output.md \
  --no-cleanup

Output location

Place output .md files in the vault at 00_inbox/ or 30_knowledge/ as appropriate. Use snake_case filenames with ISO dates, e.g. financial_times_uk_2026-03-27.md.

Dependencies

Notes

• Cleanup uses claude-haiku-4-5-20251001 — fast and cheap (~seconds per page, cents per issue).
• Change CLEANUP_MODEL in the script to claude-sonnet-4-6 for higher quality cleanup.
• Newspaper PDFs are scanned images — no text layer. OCR is required (pdftotext won't work).
• ~2-3 min per issue with cleanup enabled; ~30s without.