dev-document

Document Generation

Goal

Write the code that makes an application produce documents — PDF, DOCX, XLSX, PPTX — with the right library for the stack.

Not this skill: when the deliverable is a file Claude writes, reads or edits itself (a memo, a filled-in form, a reviewed contract with tracked changes), Anthropic's own skills do it far better: anthropics/skills — docx, pdf, xlsx, pptx (reading and editing existing files, tracked changes, form filling, OCR). They ship under Anthropic's own terms (not open source): install or point to them, never copy them.

Supported formats

Instructions per format

PDF - Generation from HTML/Markdown

# Option 1: Puppeteer (Node.js)
npm install puppeteer

# Option 2: WeasyPrint (Python, HTML/CSS -> PDF, no browser)
pip install weasyprint
weasyprint input.html output.pdf

# Option 3: pandoc (Markdown -> PDF, needs a PDF engine such as weasyprint or a LaTeX install)
pandoc input.md -o output.pdf --pdf-engine=weasyprint

// Puppeteer - HTML to PDF
import puppeteer from 'puppeteer';

async function generatePDF(htmlContent: string, outputPath: string) {
  const browser = await puppeteer.launch();
  const page = await browser.newPage();
  await page.setContent(htmlContent, { waitUntil: 'networkidle0' });
  await page.pdf({
    path: outputPath,
    format: 'A4',
    margin: { top: '20mm', right: '15mm', bottom: '20mm', left: '15mm' },
    printBackground: true,
  });
  await browser.close();
}