markitdown

Convert documents into clean Markdown with Microsoft's markitdown. It keeps structure — headings, tables, lists, links, slide notes, sheet names — instead of scraping flat text, which is what makes it good LLM input.

It extracts text that is already in the file. It does not OCR: scanned PDFs and photos come back empty (see "Image-only content" below).

Setup

markitdown --version   # expect 0.1.x or later

If it's missing, install it as an isolated tool. It needs Python 3.10+: