claude-world PDF Editor
Edit and process PDF files using nano-pdf, a natural language PDF editing tool.
Prerequisites
pip install nano-pdf
Or with pipx for isolated install:
pipx install nano-pdf
Verify:
nano-pdf --version
Commands
| Command | Description |
|---|---|
nano-pdf extract <file.pdf> |
Extract all text from a PDF |
nano-pdf extract <file.pdf> --pages 1-5 |
Extract specific pages |
nano-pdf merge <a.pdf> <b.pdf> -o output.pdf |
Merge two or more PDFs |
nano-pdf split <file.pdf> --pages 1-3 -o part1.pdf |
Split out specific pages |
nano-pdf annotate <file.pdf> --page 2 --text "Review this" |
Add text annotation |
nano-pdf fill <form.pdf> --field "Name=Alice" --field "Date=2026-03-21" |
Fill form fields |
nano-pdf convert <file.pdf> --to txt -o output.txt |
Convert to text |
nano-pdf convert <file.pdf> --to html -o output.html |
Convert to HTML |
nano-pdf compress <file.pdf> -o compressed.pdf |
Compress/optimize PDF |
nano-pdf info <file.pdf> |
Show metadata (pages, author, title) |
nano-pdf rotate <file.pdf> --pages all --degrees 90 -o rotated.pdf |
Rotate pages |
Usage Examples
Extract text for reading:
nano-pdf extract report.pdf > workspace/report-text.txt
Merge multiple PDFs:
nano-pdf merge cover.pdf chapter1.pdf chapter2.pdf -o final.pdf
Split out pages 5–10:
nano-pdf split document.pdf --pages 5-10 -o excerpt.pdf
Fill a PDF form:
nano-pdf fill application.pdf \
--field "FirstName=Alice" \
--field "LastName=Smith" \
--field "Date=2026-03-21" \
-o filled-application.pdf
Get PDF metadata:
nano-pdf info contract.pdf
Rules
- Always check if CLI is installed first:
which nano-pdfornano-pdf --version - If not installed, show
pip install nano-pdfand stop - Always verify the input file exists before running any command
- When extracting large PDFs (> 50 pages), save output to
workspace/rather than displaying inline - For merge operations, list the input files in order and confirm with the user before executing
- For fill operations, first use
nano-pdf infoto list available form fields before filling - Output files default to the same directory as input unless user specifies otherwise — prefer saving to
workspace/for intermediate results - Keep responses concise: confirm action and output path in 1-2 lines
- For large files, note the page count and file size in the confirmation