Leafwork: A Local-First PDF Toolkit That Runs in Your Browser

#webdev #javascript #privacy #opensource

Most PDF tools solve the task, but they often create a privacy problem first.

You want to merge a file, split a document, convert a page, redact a line, or add a signature. The default flow on many sites is still: upload the PDF, wait for server processing, download a new file, repeat.

I built Leafwork to make the default flow local-first instead.

Leafwork is here: https://leafworkpdf.vercel.app/

What Leafwork does today

Leafwork is a free browser PDF toolkit. The enabled tools are designed to run locally in the browser, so the core document work happens on your machine instead of being uploaded to a server.

Current tools include:

Merge PDF files
Split PDF pages
Convert PDF pages to images
Convert images to PDF
Add watermarks
Sign PDFs
Redact visible sensitive content
Rotate pages
Strip common hidden metadata
Use a browser-local sandbox workspace

Compression, PDF-to-Word, and AI summarization are intentionally marked as coming soon until the quality is good enough.

The sandbox workflow

The feature I am most excited about is the sandbox.

A lot of real PDF work is not a single clean operation. You might merge a few documents, delete some pages, rotate a scan, redact one page, export images, and then save the final result.

Most tools turn that into a chain of downloads: final.pdf, final-new.pdf, merged-final-v2.pdf, and so on.

Leafwork's sandbox is meant to feel more like a small local file explorer for PDFs. You can keep files in a session-only browser workspace, inspect pages, mark pages, drag files into tools, save outputs back into the workspace, and download only when you are done.

For now, the sandbox is session-only by design. Leave or clear the workspace and the files are gone. No account storage. No background server copy.

Why browser-local tools are useful

PDFs often contain resumes, forms, addresses, signatures, contracts, bank statements, or internal notes. Even when a tool is trustworthy, uploading sensitive files for routine edits can feel unnecessary.

Browser-side processing will not solve every complex PDF case, but for many everyday workflows it is enough. The more work that can happen locally, the less users have to think about where their documents went.

What I learned building it

The hard part was not just manipulating PDFs. It was designing the workflow around the files.

A few things became obvious quickly:

Page previews matter.
Page selection matters.
Reusing outputs matters.
Some features should stay disabled until they are actually good.
A privacy promise has to be visible in the product behavior, not only in the marketing copy.