Convert To Markdown

Document to Markdown Converter

Convert various document formats to Markdown using markitdown and pandoc.

Supported Formats

| Format | Extension | Notes | | ----------- | ----------------------- | -------------------------------- | | PDF | .pdf | Text + OCR hybrid mode available | | Word | .docx | Full support | | PowerPoint | .pptx | Extracts text from slides | | Excel | .xlsx | Converts tables | | HTML | .html, .htm | Full support | | EPUB | .epub | Via pandoc | | Images | .jpg, .png, .gif, .webp | OCR extraction | | Google Docs | URL | Public docs, exports as docx |

Usage

uvx --with markitdown python ~/.claude/skills/convert-to-markdown/scripts/convert.py INPUT

Options

| Flag | Description | | ------------------------ | ------------------------------------------------ | | -o, --output PATH | Output path (default: input.md next to original) | | --mode auto\|ocr\|text | Extraction mode (default: auto) | | --hybrid | PDF only: run both text + OCR, merge results |

Examples

# Basic conversion (saves as document.md)
uvx --with markitdown python ~/.claude/skills/convert-to-markdown/scripts/convert.py document.pdf

# Force OCR mode for scanned document
uvx --with markitdown python ~/.claude/skills/convert-to-markdown/scripts/convert.py scanned.pdf --mode ocr

# Hybrid extraction for PDFs with mixed content
uvx --with markitdown python ~/.claude/skills/convert-to-markdown/scripts/convert.py report.pdf --hybrid

# Convert Word document
uvx --with markitdown python ~/.claude/skills/convert-to-markdown/scripts/convert.py report.docx

# Convert public Google Doc
uvx --with markitdown python ~/.claude/skills/convert-to-markdown/scripts/convert.py "https://docs.google.com/document/d/DOCUMENT_ID/edit"

# Specify output location
uvx --with markitdown python ~/.claude/skills/convert-to-markdown/scripts/convert.py slides.pptx -o notes.md

Mode Details

• auto: Smart detection - uses OCR for images, text extraction for documents
• text: Pure text extraction, faster but misses image content
• ocr: Force OCR on everything, slower but catches all visual content
• hybrid (PDF only): Runs both text and OCR extraction, merges results for best coverage

Google Docs

For Google Docs conversion:

• Document must be publicly accessible (or "Anyone with link can view")
• Provide the full URL: https://docs.google.com/document/d/DOC_ID/edit
• Script exports as docx, then converts to markdown

Output

• Default: Creates filename.md next to the original file
• Use -o to specify a different location
• Prints the output path when complete

Requirements

• markitdown: Installed automatically via uvx
• pandoc: Must be installed on system (for epub, odt, rst, tex formats)

``bash brew install pandoc # macOS ``

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: mfmezger
• Source: mfmezger/aiagent_dotfiles
• License: MIT

Install and usage instructions live in the source repository linked above.