Converting DOCX and PDF to Markdown with MarkItDown

MarkItDown is a free Microsoft tool that converts documents such as DOCX and PDF into Markdown. It is useful for pulling existing documents into anything that works with plain text - static sites, note systems, or LLM pipelines.

Install it inside a Python virtual environment to keep its dependencies isolated. Create the environment and activate it:

cd ~/code
virtualenv -p python3 markitdown
cd ~/code/markitdown
source bin/activate

Install the package with pip. The [all] extra pulls in the optional dependencies for every supported format:

pip install "markitdown[all]"

Verify the installation:

markitdown --help

markitdown writes the converted Markdown to stdout, so redirect it to a file. Convert a DOCX file:

markitdown report.docx > report.md

PDF files work the same way:

markitdown manual.pdf > manual.md

To convert a whole directory, loop over the files with the shell. For DOCX files:

for f in *.docx; do
  markitdown "$f" > "${f%.docx}.md"
done

And for PDF files:

for f in *.pdf; do
  markitdown "$f" > "${f%.pdf}.md"
done

DOCX files generally convert with higher accuracy than PDFs, since the text and structure are already machine-readable. Scanned PDFs have no text layer at all, so run them through OCR first (for example with ocrmypdf) before converting. After conversion, open the generated .md files and review complex tables, images, and layouts - these are the areas most likely to need manual cleanup.