Document DOCX Skill - Quick Reference

.docx

• Prefer templates + styles over manual formatting.
• Treat

  .docx

  as the editable source; treat PDF as a release artifact.
• If distributing externally, include basic accessibility hygiene (headings, table headers, alt text).

Quick Reference

Tool Selection

• Prefer

  docxtpl

  when non-developers must edit layout/design in Word.
• Prefer

  python-docx

  for structural edits (paragraphs/tables/headers/footers) when formatting complexity is moderate.
• Prefer

  docx

  (Node.js) for server-side generation in TypeScript-heavy stacks.
• Prefer

  mammoth

  for text-first extraction or DOCX-to-HTML (best effort; may drop some layout fidelity).

Known Limits (Plan Around These)

• .doc

  (legacy) is not supported by these libraries; convert to

  .docx

  first (e.g., LibreOffice).

• python-docx

  cannot reliably create true tracked changes; use Word compare or specialized OOXML tooling.
• Tables of Contents and many fields are placeholders until opened/updated in Word.

Core Operations

Create Document (Python - python-docx)

from docx import Document
from docx.shared import Inches, Pt
from docx.enum.text import WD_ALIGN_PARAGRAPH

doc = Document()

# Title
title = doc.add_heading('Document Title', 0)
title.alignment = WD_ALIGN_PARAGRAPH.CENTER

# Paragraph with formatting
para = doc.add_paragraph()
run = para.add_run('Bold and ')
run.bold = True
run = para.add_run('italic text.')
run.italic = True

# Table
table = doc.add_table(rows=3, cols=3)
table.style = 'Table Grid'
for i, row in enumerate(table.rows):
    for j, cell in enumerate(row.cells):
        cell.text = f'Row {i+1}, Col {j+1}'

# Image
doc.add_picture('image.png', width=Inches(4))

# Save
doc.save('output.docx')

Create Document (Node.js - docx)

import { Document, Packer, Paragraph, TextRun, Table, TableRow, TableCell } from 'docx';
import * as fs from 'fs';

const doc = new Document({
  sections: [{
    properties: {},
    children: [
      new Paragraph({
        children: [
          new TextRun({ text: 'Bold text', bold: true }),
          new TextRun({ text: ' and normal text.' }),
        ],
      }),
      new Table({
        rows: [
          new TableRow({
            children: [
              new TableCell({ children: [new Paragraph('Cell 1')] }),
              new TableCell({ children: [new Paragraph('Cell 2')] }),
            ],
          }),
        ],
      }),
    ],
  }],
});

Packer.toBuffer(doc).then((buffer) => {
  fs.writeFileSync('output.docx', buffer);
});

Template-Based Generation (Python - docxtpl)

from docxtpl import DocxTemplate

doc = DocxTemplate('template.docx')
context = {
    'company_name': 'Acme Corp',
    'date': '2025-01-15',
    'items': [
        {'name': 'Widget A', 'price': 100},
        {'name': 'Widget B', 'price': 200},
    ]
}
doc.render(context)
doc.save('filled_template.docx')

Extract Content (Python - python-docx)

from docx import Document

doc = Document('input.docx')

# Extract all text
full_text = []
for para in doc.paragraphs:
    full_text.append(para.text)

# Extract tables
for table in doc.tables:
    for row in table.rows:
        row_data = [cell.text for cell in row.cells]
        print(row_data)

Styling Reference

add_heading(text, 1)

HeadingLevel.HEADING_1

run.bold = True

TextRun({ bold: true })

run.italic = True

TextRun({ italics: true })

run.font.size = Pt(12)

TextRun({ size: 24 })

WD_ALIGN_PARAGRAPH.CENTER

AlignmentType.CENTER

doc.add_page_break()

new PageBreak()

Do / Avoid (Dec 2025)

Do

• Use consistent heading levels and a table of contents for long docs.
• Capture decisions and action items with owners and due dates.
• Store docs in a versioned, searchable system.

Avoid

• Manual formatting instead of styles (breaks consistency).
• Docs with no owner or review cadence (stale quickly).
• Copy/pasting without updating definitions and links.

Output Quality Checklist

• Structure: consistent heading hierarchy, styles, and (when needed) an auto-generated table of contents.
• Decisions: decisions/actions captured with owner + due date (not buried in prose).
• Versioning: doc ID + version + change summary; review cadence defined.
• Accessibility hygiene: headings/reading order are correct; table headers are marked; alt text for non-decorative images.
• Reuse: use

  assets/doc-template-pack.md

  for decision logs and recurring doc types.

Optional: AI / Automation

• Summarize meeting notes into decisions/actions; humans verify accuracy.
• Draft first-pass docs from outlines; do not invent facts or quotes.

Navigation

• references/docx-patterns.md - Advanced formatting, styles, headers/footers
• references/template-workflows.md - Mail merge, batch generation
• references/tracked-changes.md - Tracked changes: what is feasible, and what is not
• data/sources.json - Library documentation links

• scripts/docx_inspect_ooxml.py

  - Dependency-free OOXML inspection (including tracked changes signals)

• scripts/docx_extract.py

  - Extract text/tables to JSON (requires

  python-docx

  )

• scripts/docx_render_template.py

  - Render a

  docxtpl

  template (requires

  docxtpl

  )

• scripts/docx_to_html.mjs

  - Convert

  .docx

  to HTML (requires

  mammoth

  )

• assets/report-template.md - Standard report structure
• assets/contract-template.md - Legal document structure
• assets/doc-template-pack.md - Decision log, meeting notes, changelog templates

• ../document-pdf/SKILL.md - PDF generation and conversion
• ../docs-codebase/SKILL.md - Technical writing patterns