I Built a PDF Toolbox That Never Uploads Your Files — Here Is Why

Last week I googled “compress pdf online free.” The top result was a well-known tool that processed my file in seconds. Then I read their privacy policy. It said they “may retain uploaded files for up to 24 hours for quality assurance purposes.”

My tax return was sitting on a stranger's server for 24 hours.

That was the moment I decided to build PDF Toolbox — a free online PDF tool suite that never uploads your files to any server. Everything happens right in your browser.

The Dirty Secret of Online PDF Tools

Almost every “free online PDF tool” works the same way:

1. You drag your PDF into their website
2. Your file gets uploaded to their server (often on AWS/GCP)
3. They process it server-side using libraries like Ghostscript or LibreOffice
4. You download the result
5. Your file sits on their server for some “retention period”

Here is what the privacy policies of popular tools actually say:

“Just 1 hour” or “only 24 hours” might sound harmless. But consider what people upload: contracts, tax returns, medical records, bank statements, legal documents, NDAs, passport scans. Any of these in the wrong hands is a disaster.

And it is not just about malicious intent. Servers get hacked. Employees access files they shouldn't. Law enforcement issues subpoenas. Your data is only as safe as the weakest link in their infrastructure chain.

How Browser-Based PDF Processing Works

PDF Toolbox uses three browser technologies to process PDFs entirely on your device:

1. WebAssembly (WASM)

We compile PDF manipulation libraries from C/C++ to WebAssembly. This lets us run near-native-performance code directly in your browser. The same algorithms that power desktop PDF tools now run in a sandboxed browser environment.

2. PDF.js

Mozilla's PDF.js renders PDFs in the browser. We use it to parse, render, and extract content from PDF files — all client-side. It is the same library that powers Firefox's built-in PDF viewer.

3. Web Crypto API

For password-protected PDFs, we use the browser's built-in cryptographic APIs to handle encryption and decryption — no keys ever leave your machine.

The Trade-offs

Browser-based processing is not perfect. Here is where it falls short compared to server-side tools:

• File size limits: Your device's RAM limits how large a PDF you can process. Most modern devices handle up to 100MB easily. Server-side can handle gigabyte files.
• No OCR: Optical character recognition requires large ML models (often 500MB+), which is impractical to run in a browser. We are exploring smaller models.
• No collaboration: Server-side tools enable real-time collaboration. Browser-only tools are single-user.
• Mobile performance: Large PDFs may be slow on older phones due to limited RAM and CPU.

But for 95% of PDF tasks — compress, merge, split, convert — the browser handles it perfectly.

What the Tech Stack Looks Like

For the curious developers out there:

• Frontend: Next.js 16 + TypeScript + Tailwind CSS
• PDF Processing: pdf-lib (merge, split, unlock) + browser canvas API (PDF to JPG) + custom WASM (compression)
• Deployment: Vercel (static + edge functions)
• Analytics: Vercel Analytics (privacy-respecting, no cookies)

The entire app is static — no backend server, no database, no file storage. The only server involved is the CDN that delivers the JavaScript bundle to your browser.

Is This the Future of Web Apps?

I believe so. WebAssembly is maturing fast. Browser APIs are getting more powerful. The line between “native app” and “web app” is blurring. Five years ago, you needed a server for any kind of PDF manipulation. Today, your browser can do it faster than the round-trip to a server.

This shift matters for more than just PDFs. Photo editing (Photopea), video transcoding, code compilation (StackBlitz), and even CAD modeling now run in browsers. The era of “upload to process” is coming to an end.

Try It Yourself

PDF Toolbox is completely free, open source, and requires no registration. Every tool works the same way: drop your file, process it, download the result. Nothing leaves your computer.

Built by hwlsniper — a developer who believes your files belong to you, not some server.