MyKK PDF Viewer

A fast, mobile-first, single-file PDF viewer. Drag & drop a .pdf anywhere on the page and start reading instantly. It's a viewer, not an editor — no toolbars, no accounts, no uploads. Everything runs locally in your browser.

🔗 Live: https://pdf-viewer.us/

Part of the File Viewer family — HTML, Markdown, ePUB, PDF, and Data each have their own dedicated viewer. Use the ☰ menu in the header to jump between them.

Features

• 📄 Drag & drop anywhere — drop a .pdf on the page to read it. You can also paste a copied file, or tap the page to open a file picker.
• 🏷️ Document title in the header — pulled from the PDF's metadata (with the file name as a fallback) and shown in the top-left.
• 🗂️ Outline / bookmarks — a list button in the header opens a slide-in panel built from the PDF's outline, with nested sections. Click any entry to jump straight to that page. PDFs without an outline get a simple page list.
• 📜 Continuous scroll — all pages flow in one smooth column, rendered lazily as you scroll so even large documents open quickly and stay light on memory.
• 🔍 Zoom — zoom in/out (buttons, or + / -), and tap Fit (or press 0) to snap back to fit-width. Renders crisply on high-DPI screens.
• ⏮️⏭️ Page navigation — Previous/Next buttons and a "page x / y" readout in a floating toolbar; type a page number to jump, or use ←/→, PageUp/PageDown, and Home/End.
• 🔤 Selectable text — a real text layer sits over every page, so you can select and copy text just like in a desktop reader.
• 📋 Copy — copies your current text selection, or the current page's text if nothing is selected.
• 🗑️ Clear — closes the PDF and empties the view.
• 🔑 Password-protected PDFs — you're prompted for the password when a PDF is encrypted.
• 🎨 Pick any background color — a color swatch in the header tints the app and the gutter around the pages (the pages themselves stay as white "paper"). The UI adapts for contrast, and your choice is remembered — saved to a cookie, with a localStorage fallback so even a downloaded file:// copy remembers it. Starts on white.
• 🫥 Distraction-free — while you read, the header hides when you scroll down and comes back when you scroll up (or after a few seconds of stillness).
• 📱 Mobile-first & responsive, with safe-area support for notched phones.
• 🔒 Safe by default — your files are parsed and rendered entirely in the browser and never leave your device.
• 🪶 One file, no build — index.html is self-contained and the viewer works offline, even straight from file://.
• 📊 Privacy-friendly analytics — the hosted site uses a self-hosted, cookieless Plausible instance (see Privacy).

Quick start

Just open it. Download index.html, double-click it, and you're done — it runs locally with no server, no build, and no internet connection.

Prefer a local server?

python3 -m http.server 8080   # then open http://localhost:8080

Any static web server works too — there is nothing to build.

Deploy to Cloudflare Pages (auto-deploy from this repo)

Connect the repository in the Cloudflare dashboard (Workers & Pages → Create → Pages → Connect to Git) and use:

Setting	Value
Production branch	main
Framework preset	None
Build command	(leave blank)
Build output directory	/

It's a static site — Cloudflare just serves index.html. The included _headers file applies security headers (CSP, nosniff, frame-ancestors, etc.) automatically. Every push to main then auto-deploys.

Add your custom domain pdf-viewer.us under the project's Custom domains tab — Cloudflare creates the DNS record and provisions SSL for you.

How it works

Everything is in index.html:

• Your HTML/CSS/UI and the app logic — no build step, no tooling.
• PDF.js v3.11.174 (Apache-2.0) parses and renders each page to a <canvas> and produces the selectable text layer, entirely in memory.
• PDF.js runs in a Web Worker so parsing and rendering stay off the main thread. The worker is created from an inline blob: URL, so there are no network requests — and if a browser blocks blob workers, PDF.js transparently falls back to running on the main thread (so it still works from file://).
• Pages are rendered lazily with an IntersectionObserver: only pages near the viewport hold a canvas, so memory stays bounded on long documents.

The library is embedded inline (minified, with its license banner retained), so the viewer renders with no network dependencies and works fully offline. The only external request is the analytics script on the hosted site (see Privacy).

Note: because it bundles the full PDF.js engine, index.html is larger than a plain-text viewer (~1.4 MB). That's the price of rendering real PDFs with zero network dependencies — and it's still a single file with no build.

Limitations

• Interactive PDF forms and annotations (widgets, comments) are not rendered — this is a reader for viewing and reading, not filling forms.
• Scanned PDFs with no embedded text have nothing to select or copy (there's no OCR).
• Very large or complex PDFs render page-by-page as you scroll; the first pages appear quickly while the rest render on demand.

Privacy

Your files stay on your device — they're parsed and rendered in the browser and never uploaded anywhere, and your content is never sent to any server.

The hosted site at pdf-viewer.us uses a self-hosted Plausible instance for privacy-friendly, cookieless page analytics: aggregate visit counts only — no personal data, no cross-site tracking, no advertising. Plausible automatically ignores localhost and file://, so a downloaded local copy reports nothing.

The only thing stored on your device is your chosen background color (a cookie, with a localStorage fallback).

License

MIT © 2026 Michal Ferber, aka TechGuyWithABeard.

The bundled third-party library retains its own license (PDF.js: Apache-2.0) — see LICENSE for details.