DOCX Skill

Description

This skill enables creating and manipulating .docx Word documents programmatically using the docx npm library. Supports paragraphs, headings, tables, images, lists, page breaks, headers/footers, and styles.

Instructions

When the user asks you to create or edit Word documents, use these patterns:

Creating a Basic Document

Copy
const docx = require('docx');
const fs = require('fs');

const doc = new docx.Document({
  sections: [{
    properties: {},
    children: [
      new docx.Paragraph({
        children: [
          new docx.TextRun({ text: 'Project Report', bold: true, size: 48 }),
        ],
        heading: docx.HeadingLevel.HEADING_1,
      }),
      new docx.Paragraph({
        children: [
          new docx.TextRun('Generated on: ' + new Date().toLocaleDateString()),
        ],
        spacing: { after: 200 },
      }),
    ],
  }],
});

const buffer = await docx.Packer.toBuffer(doc);
fs.writeFileSync('report.docx', buffer);

Adding Tables

Copy
const table = new docx.Table({
  rows: [
    new docx.TableRow({
      children: [
        new docx.TableCell({
          children: [new docx.Paragraph('Name')],
          shading: { fill: '4472C4' },
        }),
        new docx.TableCell({
          children: [new docx.Paragraph('Role')],
          shading: { fill: '4472C4' },
        }),
      ],
      tableHeader: true,
    }),
    new docx.TableRow({
      children: [
        new docx.TableCell({ children: [new docx.Paragraph('Alice')] }),
        new docx.TableCell({ children: [new docx.Paragraph('Engineer')] }),
      ],
    }),
  ],
  width: { size: 100, type: docx.WidthType.PERCENTAGE },
});

Adding Images

Copy
const imageBuffer = fs.readFileSync('logo.png');
const image = new docx.ImageRun({
  data: imageBuffer,
  transformation: { width: 200, height: 100 },
  type: 'png',
});

new docx.Paragraph({ children: [image] });

Bullet Lists

Copy
const items = ['First item', 'Second item', 'Third item'];
const bullets = items.map(text => new docx.Paragraph({
  children: [new docx.TextRun(text)],
  bullet: { level: 0 },
}));

Rules

• Install dependency first: npm install docx
• The docx library creates new documents only — it cannot read/parse existing .docx files
• To read existing .docx files, use mammoth (npm install mammoth) for text/HTML extraction
• Always use Packer.toBuffer() for writing (async operation)
• Images must be loaded as Buffer, not file paths
• Use HeadingLevel.HEADING_1 through HEADING_6 for proper document structure
• Set page margins, orientation, and size via section properties
• For complex layouts, use multiple sections with different properties

Examples

User: Create a project proposal document Action: Generate a formatted .docx with title page, sections, and table of deliverables

User: Convert this markdown to a Word doc Action: Parse the markdown structure, create equivalent docx elements

User: Generate invoices from this data Action: Create a template with header, line-item table, and totals