Setup

• If this fails (command not found), install Python 3.10 or newer:
  • Windows: winget install --id Python.Python.3.12 -e
  • macOS: brew install python@3.12
  • Linux (Debian/Ubuntu): sudo apt-get update && sudo apt-get install -y python3 python3-pip python-is-python3
• If the reported version is older than 3.10, install a newer Python using the same command above (MarkItDown requires 3.10+).

2. Check pip is available

• If this fails, bootstrap pip:

3. Install MarkItDown with PDF support, plus PyMuPDF for image extraction

Use the scripts/requirements.txt file bundled with this skill to install pinned, known-good versions of the dependencies:

This pulls in markitdown[pdf] and pymupdf>=1.24.0. PyMuPDF (imported as fitz) is required separately because MarkItDown’s PDF converter only extracts text and tables — it has no support for embedded images at all, so this skill’s script extracts them itself.

4. Verify the install

Expect to see markitdown + pymupdf OK printed with no errors. If you see a ModuleNotFoundError, repeat step 3 — pip may be installing into a different Python environment than the one being invoked (check python -m pip --version shows the same path as python --version‘s interpreter).

Notes

• This setup only needs to be done once per environment/virtual environment, not once per conversion.
• convert_pdf_to_md.py itself also checks for markitdown and fitz at startup and prints a pointer back to this file if either is missing, so re-running setup is safe and idempotent.
• Only .pdf is supported by this skill — it’s MarkItDown’s only PDF-family format, so there’s no legacy-format equivalent to worry about (unlike Word’s .doc or Excel’s .xls).
• Scanned/image-only PDFs (no embedded text layer) will produce little or no text from MarkItDown, since it does not perform OCR. The images themselves will still be extracted and appended, but the text body may be empty or near-empty in that case — mention this to the user if it happens.