MarkItDown

Convert files and URLs to Markdown via the markitdown CLI.

Usage

# File to stdout
markitdown <file>

# File to output file
markitdown <file> -o output.md

# Stdin with extension hint
cat file.pdf | markitdown -x .pdf

# With MIME type hint
markitdown <file> -m application/pdf

Supported Formats

Format	Extensions	Notes
PDF	.pdf	Text extraction with layout preservation
Word	.docx	Full formatting, tables, images
Excel	.xlsx, .xls	Each sheet as a Markdown table
PowerPoint	.pptx	Slides with text and notes
HTML	.html, .htm	Converts to clean Markdown
CSV	.csv	Renders as Markdown table
JSON/XML	.json, .xml	Structured text extraction
EPUB	.epub	E-book to Markdown
Images	.jpg, .png, .gif, etc.	EXIF metadata extraction
Outlook	.msg	Email headers, body, attachments
Jupyter	.ipynb	Cells rendered as Markdown + code blocks
RSS/Atom	feeds	Feed entries with titles, links, content
ZIP	.zip	Recursively converts contained files
YouTube	URL	Transcript extraction
Wikipedia	URL	Article content
Plain text	.txt, .md, .py, etc.	Pass-through with charset handling

Key Options

• -o FILE — Write output to file instead of stdout
• -x EXT — Hint file extension (for stdin: -x .pdf)
• -m MIME — Hint MIME type
• -c CHARSET — Hint charset (e.g., UTF-8)

Tips

• Redirect stderr to suppress noisy onnxruntime/pydub warnings: markitdown file.pdf 2>/dev/null
• For large files, pipe to head to preview: markitdown big.xlsx 2>/dev/null | head -100
• ZIP files are recursively converted — each contained file gets its own section
• Excel workbooks produce one table per sheet
• For URLs (YouTube, Wikipedia), pass the URL directly as the filename argument