Skills
▥NYC
skills/smithery/ai/office-to-md

office-to-md

SKILL.md

Office to Markdown Skill

Overview

This skill enables conversion from various Office formats to Markdown using markitdown - Microsoft's open-source tool for converting documents to Markdown. Perfect for making Office content searchable, version-controllable, and AI-friendly.

How to Use

  1. Provide the Office file (Word, Excel, PowerPoint, PDF, etc.)
  2. Optionally specify conversion options
  3. I'll convert it to clean Markdown

Example prompts:

  • "Convert this Word document to Markdown"
  • "Turn this PowerPoint into Markdown notes"
  • "Extract content from this PDF as Markdown"
  • "Convert this Excel file to Markdown tables"

Domain Knowledge

markitdown Fundamentals

from markitdown import MarkItDown

# Initialize converter
md = MarkItDown()

# Convert file
result = md.convert("document.docx")
print(result.text_content)

# Save to file
with open("output.md", "w") as f:
    f.write(result.text_content)

Supported Formats

Format Extension Notes
Word .docx Full text, tables, basic formatting
Excel .xlsx Converts to Markdown tables
PowerPoint .pptx Slides as sections
PDF .pdf Text extraction
HTML .html Clean markdown
Images .jpg, .png OCR with vision model
Audio .mp3, .wav Transcription
ZIP .zip Processes contained files

Basic Usage

Python API

from markitdown import MarkItDown

# Simple conversion
md = MarkItDown()
result = md.convert("document.docx")

# Access content
markdown_text = result.text_content

# With options
md = MarkItDown(
    llm_client=None,      # Optional LLM for enhanced processing
    llm_model=None        # Model name if using LLM
)

Command Line

# Install
pip install markitdown

# Convert file
markitdown document.docx > output.md

# Or with output file
markitdown document.docx -o output.md

Word Document Conversion

from markitdown import MarkItDown

md = MarkItDown()

# Convert Word document
result = md.convert("report.docx")

# Output preserves:
# - Headings (as # headers)
# - Bold/italic formatting
# - Lists (bulleted and numbered)
# - Tables (as markdown tables)
# - Hyperlinks

print(result.text_content)

Example Output:

# Annual Report 2024

## Executive Summary

This report summarizes the key achievements and challenges...

### Key Metrics

| Metric | 2023 | 2024 | Change |
|--------|------|------|--------|
| Revenue | $10M | $12M | +20% |
| Users | 50K | 75K | +50% |

## Detailed Analysis

The following sections provide...

Excel Conversion

from markitdown import MarkItDown

md = MarkItDown()
result = md.convert("data.xlsx")

# Each sheet becomes a section
# Data becomes markdown tables
print(result.text_content)

Example Output:

## Sheet1

| Name | Department | Salary |
|------|------------|--------|
| John | Engineering | $80,000 |
| Jane | Marketing | $75,000 |

## Sheet2

| Product | Q1 | Q2 | Q3 | Q4 |
|---------|----|----|----|----|
| Widget A | 100 | 120 | 150 | 180 |

PowerPoint Conversion

from markitdown import MarkItDown

md = MarkItDown()
result = md.convert("presentation.pptx")

# Each slide becomes a section
# Speaker notes included if present
print(result.text_content)

Example Output:

# Slide 1: Company Overview

Our mission is to...

## Key Points
- Innovation first
- Customer focused
- Global reach

---

# Slide 2: Market Analysis

The market opportunity is significant...

**Notes:** Mention the competitor analysis here

PDF Conversion

from markitdown import MarkItDown

md = MarkItDown()
result = md.convert("document.pdf")

# Extracts text content
# Tables converted where detected
print(result.text_content)

Image Conversion (with Vision Model)

from markitdown import MarkItDown
import anthropic

# Use Claude for image description
client = anthropic.Anthropic()

md = MarkItDown(
    llm_client=client,
    llm_model="claude-sonnet-4-20250514"
)

result = md.convert("diagram.png")
print(result.text_content)

# Output: Description of the image content

Batch Conversion

from markitdown import MarkItDown
from pathlib import Path

def batch_convert(input_dir, output_dir):
    """Convert all Office files to Markdown."""
    md = MarkItDown()
    input_path = Path(input_dir)
    output_path = Path(output_dir)
    output_path.mkdir(exist_ok=True)
    
    extensions = ['.docx', '.xlsx', '.pptx', '.pdf']
    
    for ext in extensions:
        for file in input_path.glob(f'*{ext}'):
            try:
                result = md.convert(str(file))
                output_file = output_path / f"{file.stem}.md"
                
                with open(output_file, 'w') as f:
                    f.write(result.text_content)
                
                print(f"Converted: {file.name}")
            except Exception as e:
                print(f"Error converting {file.name}: {e}")

batch_convert('./documents', './markdown')

Best Practices

  1. Check Output Quality: Review converted Markdown for accuracy
  2. Handle Tables: Complex tables may need manual adjustment
  3. Preserve Structure: Use consistent heading levels in source docs
  4. Image Handling: Consider using vision models for important images
  5. Version Control: Store converted Markdown in Git for tracking

Common Patterns

Document Archive

import os
from datetime import datetime
from markitdown import MarkItDown

def archive_document(doc_path, archive_dir):
    """Convert and archive Office document to Markdown."""
    md = MarkItDown()
    result = md.convert(doc_path)
    
    # Create archive structure
    date_str = datetime.now().strftime('%Y-%m-%d')
    filename = os.path.basename(doc_path)
    base_name = os.path.splitext(filename)[0]
    
    # Save with metadata
    output_content = f"""---
source: {filename}
converted: {date_str}
---

{result.text_content}
"""
    
    output_path = os.path.join(archive_dir, f"{base_name}.md")
    with open(output_path, 'w') as f:
        f.write(output_content)
    
    return output_path

AI-Ready Corpus

from markitdown import MarkItDown
from pathlib import Path
import json

def create_ai_corpus(doc_folder, output_file):
    """Convert documents to JSON corpus for AI training/RAG."""
    md = MarkItDown()
    corpus = []
    
    for doc in Path(doc_folder).glob('**/*'):
        if doc.suffix in ['.docx', '.pdf', '.pptx', '.xlsx']:
            try:
                result = md.convert(str(doc))
                corpus.append({
                    'source': str(doc),
                    'filename': doc.name,
                    'content': result.text_content,
                    'type': doc.suffix[1:]
                })
            except Exception as e:
                print(f"Skipped {doc.name}: {e}")
    
    with open(output_file, 'w') as f:
        json.dump(corpus, f, indent=2)
    
    print(f"Created corpus with {len(corpus)} documents")
    return corpus

Examples

Example 1: Convert Documentation Suite

from markitdown import MarkItDown
from pathlib import Path

def convert_docs_to_wiki(docs_folder, wiki_folder):
    """Convert all Office docs to markdown wiki structure."""
    md = MarkItDown()
    docs_path = Path(docs_folder)
    wiki_path = Path(wiki_folder)
    
    # Create wiki structure
    wiki_path.mkdir(exist_ok=True)
    
    # Create index
    index_content = "# Documentation Index\n\n"
    
    for doc in sorted(docs_path.glob('**/*.docx')):
        try:
            result = md.convert(str(doc))
            
            # Create relative path in wiki
            rel_path = doc.relative_to(docs_path)
            output_file = wiki_path / rel_path.with_suffix('.md')
            output_file.parent.mkdir(parents=True, exist_ok=True)
            
            # Write markdown
            with open(output_file, 'w') as f:
                f.write(result.text_content)
            
            # Add to index
            link = str(rel_path.with_suffix('.md')).replace('\\', '/')
            index_content += f"- [{doc.stem}]({link})\n"
            
            print(f"Converted: {doc.name}")
            
        except Exception as e:
            print(f"Error: {doc.name} - {e}")
    
    # Write index
    with open(wiki_path / 'index.md', 'w') as f:
        f.write(index_content)

convert_docs_to_wiki('./company_docs', './wiki')

Example 2: Meeting Notes Processor

from markitdown import MarkItDown
import re
from datetime import datetime

def process_meeting_notes(pptx_path):
    """Extract and structure meeting notes from PowerPoint."""
    md = MarkItDown()
    result = md.convert(pptx_path)
    
    # Parse the markdown
    content = result.text_content
    
    # Extract sections
    sections = {
        'attendees': [],
        'agenda': [],
        'decisions': [],
        'action_items': []
    }
    
    current_section = None
    
    for line in content.split('\n'):
        line_lower = line.lower()
        
        if 'attendee' in line_lower or 'participant' in line_lower:
            current_section = 'attendees'
        elif 'agenda' in line_lower:
            current_section = 'agenda'
        elif 'decision' in line_lower:
            current_section = 'decisions'
        elif 'action' in line_lower:
            current_section = 'action_items'
        elif line.strip().startswith(('-', '*', '•')) and current_section:
            sections[current_section].append(line.strip()[1:].strip())
    
    # Generate structured output
    output = f"""# Meeting Notes

**Date:** {datetime.now().strftime('%Y-%m-%d')}
**Source:** {pptx_path}

## Attendees
{chr(10).join('- ' + a for a in sections['attendees'])}

## Agenda
{chr(10).join('- ' + a for a in sections['agenda'])}

## Decisions Made
{chr(10).join('- ' + d for d in sections['decisions'])}

## Action Items
{chr(10).join('- [ ] ' + a for a in sections['action_items'])}
"""
    
    return output

notes = process_meeting_notes('team_meeting.pptx')
print(notes)

Example 3: Excel to Documentation

from markitdown import MarkItDown

def excel_to_data_dictionary(xlsx_path):
    """Convert Excel data model to data dictionary documentation."""
    md = MarkItDown()
    result = md.convert(xlsx_path)
    
    # Add documentation structure
    doc = f"""# Data Dictionary

Generated from: `{xlsx_path}`

{result.text_content}

## Usage Notes

- All tables are derived from the source Excel file
- Review data types and constraints before use
- Contact data team for clarifications

## Change Log

| Date | Change | Author |
|------|--------|--------|
| {datetime.now().strftime('%Y-%m-%d')} | Initial generation | Auto |
"""
    
    return doc

documentation = excel_to_data_dictionary('data_model.xlsx')
with open('data_dictionary.md', 'w') as f:
    f.write(documentation)

Limitations

  • Complex formatting may be simplified
  • Images are not embedded (use vision model for descriptions)
  • Some table structures may not convert perfectly
  • Track changes in Word are not preserved
  • Comments may not be extracted

Installation

pip install markitdown

# For image/audio processing
pip install markitdown[all]

# For specific features
pip install markitdown[images]  # Image OCR
pip install markitdown[audio]   # Audio transcription

Resources

Weekly Installs
1
Repository
smithery/ai
First Seen
8 days ago
Installed on
cursor1