Lesson 17 — OpenClaw Nano PDF Skill: Merge, Split, and Compress PDFs with Natural Language Commands (2026)

Goal: Install the Nano PDF Skill so OpenClaw can process PDFs with natural language: merge, split, extract pages, compress, add watermarks — all executed locally with no uploads.

What Is the Nano PDF Skill?

Local processing with no cloud uploads is Nano PDF Skill's biggest selling point. Online PDF tools (Smallpdf, iLovePDF, etc.) all require uploading your file to their servers, creating data privacy risks.

Nano PDF Skill runs on a local CLI tool — all processing happens on your device and files never leave your machine. Supported operations:

ClawHub's 7th most popular Skill (78k downloads), authored by Peter Steinberger (@steipete), a well-known iOS developer.

Step 1: Install the Dependency CLI Tool

Nano PDF Skill depends on the nano-pdf CLI. Install it first:

npm install -g nano-pdf

Verify the installation:

nano-pdf --version
# Should output a version number like 2.1.0

If you get a permission error, use:

sudo npm install -g nano-pdf

Or switch to a Node.js environment managed by nvm:

nvm use 20
npm install -g nano-pdf

Step 2: Install the Skill

In WebChat or Telegram, send:

/install @steipete/nano-pdf

Verify installation:

pnpm openclaw skills list
# nano-pdf should appear in the list

Step 3: Merge Multiple PDFs

The most commonly used operation — just describe it in natural language:

Merge ~/Documents/chapter1.pdf, ~/Documents/chapter2.pdf, and ~/Documents/chapter3.pdf into one file, save to the desktop named complete.pdf

AI will call the following command:

nano-pdf merge \
  ~/Documents/chapter1.pdf \
  ~/Documents/chapter2.pdf \
  ~/Documents/chapter3.pdf \
  -o ~/Desktop/complete.pdf

You can also specify order and add bookmarks:

Merge these three PDFs and create a bookmark for each original file for easy navigation:
~/Documents/chapter1.pdf ~/Documents/chapter2.pdf ~/Documents/chapter3.pdf

Step 4: Split a PDF (Extract by Page Range)

Flexible page selection is supported:

Extract pages 5 to 20 from ~/Downloads/report.pdf and save as summary-section.pdf

nano-pdf extract ~/Downloads/report.pdf --pages 5-20 -o ~/Desktop/summary-section.pdf

Extract non-consecutive pages:

Extract page 1, page 3, and pages 7–10 from ~/Downloads/contract.pdf and save them together

nano-pdf extract ~/Downloads/contract.pdf --pages 1,3,7-10 -o ~/Desktop/key-clauses.pdf

Split a 100-page PDF into 10-page chunks:

Split ~/Downloads/manual.pdf into separate files every 10 pages, save to ~/Desktop/manual-sections/ directory

Step 5: Compress a PDF

Four quality levels are supported:

Compress ~/Downloads/presentation.pdf at high quality, save to the same directory

nano-pdf compress ~/Downloads/presentation.pdf --quality high -o ~/Downloads/presentation_compressed.pdf

Quality level comparison:

Step 6: Add Watermarks

Both text and image watermarks are supported:

Text watermark:

Add a "Confidential" watermark to ~/Documents/quote.pdf, gray and semi-transparent, angled 45 degrees, on every page

nano-pdf watermark ~/Documents/quote.pdf \
  --text "Confidential" \
  --color gray \
  --opacity 0.3 \
  --angle 45 \
  -o ~/Documents/quote_watermarked.pdf

Image watermark (e.g., company logo):

Add the company logo watermark ~/Assets/logo.png to the bottom-right corner of every page in ~/Documents/contract.pdf

nano-pdf watermark ~/Documents/contract.pdf \
  --image ~/Assets/logo.png \
  --position bottom-right \
  --opacity 0.5 \
  -o ~/Documents/contract_watermarked.pdf

Step 7: Extract Text from a PDF

Export all text from a PDF as plain text for further processing:

Read all the text content from ~/Downloads/annual-report.pdf and format it as Markdown

nano-pdf extract-text ~/Downloads/annual-report.pdf -o ~/Desktop/annual-report-content.md

After extraction, you can have AI analyze it directly:

Extract all text from ~/Downloads/contract.pdf and find the breach of contract clauses

Step 8: Batch Process Multiple Files

Process multiple files at once — no need to do them one by one:

Compress all PDF files in ~/Downloads/invoices/ directory, overwrite originals

nano-pdf compress ~/Downloads/invoices/*.pdf --quality medium --overwrite

Add a "Draft" watermark to all PDFs on the desktop

nano-pdf watermark ~/Desktop/*.pdf --text "Draft" --opacity 0.2 --overwrite

FAQ

Can OpenClaw handle password-protected PDFs?

Yes, but you need to provide the password. Tell AI "this PDF has a password, it's 12345" and the Skill will include the password parameter in the command: nano-pdf merge file.pdf --password 12345. To remove password protection from a PDF (for files you have legitimate access to), say "remove the password protection from this PDF, I have permission, password is xxx."

What's the difference between Nano PDF and just having AI read a PDF?

Asking AI to read a PDF is about "understanding content" — AI treats the PDF as text for analysis, and the output is a text answer. Nano PDF is about "operating on files" — the output is a new PDF file with all structure, formatting, images, and layout preserved. The former is for Q&A analysis; the latter is for file processing. They work together well: use Nano PDF to extract specific pages, then have AI analyze the extracted content.

Will it choke on large files (100MB+)?

It won't crash, but processing time increases. nano-pdf CLI runs on your local CPU — a 100MB PDF compression typically takes 10–30 seconds. For batch processing, consider adding "run in background and notify me when done" to your instruction, and Proactive Agent will send a push notification when processing completes rather than making you wait. For very large files (500MB+), consider splitting them into smaller chunks first.

Where are processed PDFs saved?

By default, they're saved to the path you specify in your instruction. If you don't specify a path, the Skill will ask where you want to save it, or default to the same directory as the original file with an operation-type suffix (e.g., report_compressed.pdf, report_merged.pdf). All results stay local — nothing is uploaded to any cloud service.

Next Steps

• Lesson 18 — Install the Humanizer Skill to remove AI writing patterns from your articles in 5 steps