MarkItDown

What it does

MarkItDown converts many file types (PDF, Word, PowerPoint, Excel and more) to Markdown that keeps the structure: headings, lists, tables and links. That is the format LLMs read best.

Why it is useful

• One tool for most of the documents you want to feed to a model.
• Markdown is compact, so you spend fewer tokens than with raw HTML.

How to use it

pip install 'markitdown[all]'
markitdown report.pdf > report.md

Watch out for

• It reads files with the permissions of your user. The README warns you to sanitize untrusted input.
• Scanned PDFs need the separate OCR add-on.