pdf-skill

PDF Skill

Purpose

Provides expertise in programmatic PDF generation, parsing, and manipulation. Specializes in creating PDFs from scratch, extracting content, merging/splitting documents, and handling forms using PDFKit, PDF.js, Puppeteer, and similar tools.

When to Use

• Generating PDFs programmatically
• Extracting text or data from PDFs
• Merging or splitting PDF documents
• Filling PDF forms programmatically
• Converting HTML to PDF
• Adding watermarks or annotations
• Parsing PDF structure and metadata
• Building PDF report generators

Quick Start

Invoke this skill when:

• Generating PDFs from code or data
• Extracting content from PDF files
• Merging, splitting, or manipulating PDFs
• Filling or creating PDF forms
• Converting HTML/web pages to PDF

Do NOT invoke when:

• Word document creation → use /docx-skill
• Excel/spreadsheet work → use /xlsx-skill
• PowerPoint creation → use /pptx-skill
• General file operations → use Bash or file tools

Decision Framework

PDF Operation?
├── Generate from scratch
│   ├── Simple → PDFKit (Node) / ReportLab (Python)
│   └── Complex layouts → Puppeteer/Playwright + HTML
├── Parse/Extract
│   ├── Text extraction → pdf-parse / PyPDF2
│   └── Table extraction → Camelot / Tabula
├── Manipulate
│   └── pdf-lib (merge, split, edit)
└── Forms
    └── pdf-lib (fill) / PDFtk (advanced)

Core Workflows

1. PDF Generation with PDFKit

1. Install PDFKit (npm install pdfkit)
2. Create new PDDocument
3. Add content (text, images, graphics)
4. Style with fonts and colors
5. Add pages as needed
6. Pipe to file or response

2. HTML to PDF Conversion

1. Set up Puppeteer/Playwright
2. Navigate to HTML content or URL
3. Configure page size and margins
4. Set print options (headers, footers)
5. Generate PDF buffer
6. Save or stream result

3. PDF Parsing and Extraction

1. Choose parser (pdf-parse, PyPDF2, pdfplumber)
2. Load PDF file
3. Extract text or structured data
4. Handle multi-page documents
5. Clean and normalize extracted text
6. Output in desired format

Best Practices

• Use vector graphics over raster when possible
• Embed fonts for consistent rendering
• Test PDF output across different readers
• Handle large PDFs with streaming
• Use appropriate library for task complexity
• Consider accessibility (tagged PDFs)

Anti-Patterns

Anti-Pattern	Problem	Correct Approach
Image-only PDFs	Not searchable/accessible	Use text with fonts
No font embedding	Rendering issues	Embed required fonts
Memory loading large PDFs	Crashes	Stream processing
Ignoring encryption	Security/access issues	Handle encrypted PDFs
Wrong tool for job	Over-engineering	Match tool to complexity