PDF Translator Skill

PDFs are still the dominant carrier for papers and technical specs, but they’re hostile to downstream workflows: locked layout, mixed encodings, no easy way to post-process translated output. This skill closes that gap by letting Claude extract text from a PDF, translate it into a target language, and write back a clean Markdown file you can keep iterating on.

Structure

  • SKILL.md: The main definition file for the skill.
  • requirements.txt: Python dependencies.
  • references/: Reference documentation.
    • api_guide.md: API usage guide and examples.
  • scripts/: Helper scripts.
    • extract_text.py: Extracts text from a PDF file using PyPDF2.
    • generate_md.py: (Optional) Helper to save translated content with a metadata header.
    • create_test_pdf.py: Utility to generate a sample PDF for testing.
  • test_sample.pdf: Sample PDF for testing purposes.
  • test_output.md: Example output of a translated PDF.

Setup

  1. Ensure you have Python 3 installed.

  2. Install dependencies:

    pip install -r requirements.txt

Usage

You can ask Claude to translate a PDF file naturally.

Example: “Translate the file documents/paper.pdf to Spanish.”

Claude will:

  1. Read the PDF using extract_text.py.
  2. Translate the content.
  3. Save it as documents/paper_translated.md.