OfflinePDF is a privacy-first set of PDF tools that run entirely in your browser. Files are never uploaded — there are no accounts, no file limits tied to a subscription, and no watermarks.
Live site: hosted on Vercel (set NEXT_PUBLIC_SITE_URL to your production domain).
Most free PDF websites work by uploading your document to a server, processing it there, and sending it back. OfflinePDF does the opposite — every tool runs locally, in your browser, using the device's own processing power. Close the tab and the file is gone; nothing was ever sent anywhere.
- Works offline after your first visit. The site saves what it needs to run — the pages, the app code, and the tools' processing engines — so it keeps working without an internet connection. This applies to every tool except P2P Share and Whiteboard, which need a live connection by design and will show an "unavailable offline" message. The files needed for offline use finish saving a few seconds after your first visit, so give it a moment before going offline.
- 42 tools, organized into five groups: essentials, edit & organize, security, convert, and capture & share.
- Size limits by tool type — lighter tools allow files up to 300 MB; heavier tools up to 150 MB; the OCR tool allows up to 75 MB or 75 pages.
- Search and voice search in the header to quickly find the right tool.
| Category | Tools |
|---|---|
| Essentials | Merge PDF, Split PDF, Compress PDF |
| Edit & Organize | Organize Pages, Add Watermark, Extract Text, Rotate PDF, Crop & Resize, Page Numbers, Headers & Footers, OCR PDF, Compare PDFs, Repair PDF, Edit PDF Text |
| Security | Encrypt PDF, Remove Password, Flatten PDF, Redact PDF, Privacy Scanner, Fingerprint PDF |
| Convert | PDF to JPG, Images to PDF, Word to PDF, Invert Colours, PDF to ZIP, Markdown to PDF, HTML to PDF, CSV to PDF, Excel to PDF, PDF to Word, Create PDF, PDF to EPUB, PowerPoint to PDF, PDF to PowerPoint, PDF to Excel, PDF to HTML, eBook to PDF, PDF to Audio (listen only) |
| Capture, Share & Create | POS Billing, Scan to PDF, P2P Share*, Collaborative Whiteboard* |
* P2P Share and Whiteboard connect two browsers directly to each other for a live session. Files and drawings pass straight between the two people using them and are never stored on any server.
Step-by-step guides for each tool are available under /blog.
Twelve of these tools — merge, split, rotate, organize pages, watermark, page numbers, flatten, headers/footers, crop & resize, fingerprint, scan/strip metadata, and CSV to PDF — are also published as a standalone package for other developers to use in their own projects: offlinepdf-sdk.
npm install offlinepdf-sdkIt lives in this repo at packages/offlinepdf-sdk, and the website itself uses it for those 12 tools — so there's a single, shared implementation rather than two copies to keep in sync. See the usage guide for help choosing the right function, or that package's README for the full reference and what isn't included yet.
The SDK's 12 tools, plus four more (extract text, OCR, repair, and a simplified receipt generator — 17 in total), are also available as a local server that Claude Desktop or Claude Code can call directly: offlinepdf-mcp. It gives Claude the same tools as the website, with the same guarantee that nothing leaves your device — including the OCR language model, which is included in the package rather than downloaded separately.
It lives in this repo at packages/offlinepdf-mcp. It isn't published anywhere for direct install — see that package's README for setup instructions, the full tool list, and what's intentionally left out.
| Layer | Stack |
|---|---|
| App | Next.js 16 (App Router, static export), React 19, TypeScript |
| UI | Tailwind CSS 4, @tailwindcss/typography |
| pdf-lib, pdf.js | |
| Encrypt | qpdf-wasm (AES-256; extra security headers apply only on encrypt/unlock pages — see vercel.json) |
| OCR | tesseract.js (English model included; nothing downloaded at runtime) |
| Word | mammoth (Word to PDF), docx (PDF to Word) |
| PowerPoint | pptxgenjs (PDF to PowerPoint), fflate XML parser (PowerPoint to PDF) |
| Spreadsheets | SheetJS (xlsx v0.20.3, loaded from its own CDN) for Excel to PDF and PDF to Excel |
| HTML / Markdown | DOMPurify, marked |
| Zip | fflate |
| Live sharing | PeerJS for direct browser-to-browser connections, Google's connection-brokering service (p2p-share, whiteboard) |
| Camera & Audio | the browser's own camera access (scan-to-pdf) and text-to-speech (pdf-to-audio) |
| QR Codes | qrcode-generator (p2p-share, whiteboard) |
| Hosting | Vercel (deploys automatically from main) |
All processing runs through a single background worker (lib/workers/pdf.worker.ts) that loads the code for each tool only when it's needed.
npm install
npm run dev # http://localhost:3000
npm run build # static site → out/Note: the local dev server does not apply the production security headers defined in
vercel.json. Encrypt PDF needs those headers to work, so usenpm run build && npm run e2e(or a Vercel preview) to test it end-to-end.
| Script | Purpose |
|---|---|
npm run lint |
Code style check |
npm run check:limits |
Verifies the file-size limits are configured correctly |
npm run check:engines |
Quick smoke test of each tool's processing engine |
npm run build |
Production build (static export) |
npm run check-links |
Confirms every internal link in the built site resolves |
npm run e2e |
Full browser test suite against the production build; confirms no outside network requests are made |
npm run ci |
Runs lint, limits, engines, build, and link checks in sequence |
node scripts/capture-screenshot.mjs |
Captures a fresh homepage screenshot for this README |
GitHub Actions (.github/workflows/ci.yml) runs on every push and pull request to main:
- Install dependencies
- Lint and basic checks (
check:limits,check:engines) - Production build
- Link check on the built site
- End-to-end tests (Chromium)
One setup requirement: installing dependencies needs access to https://cdn.sheetjs.com. The spreadsheet library this project depends on (xlsx, version 0.20.3) is only distributed from that address — the copy on the regular npm registry is an older version with known security issues. This applies to local setup, GitHub Actions, and the Vercel build alike. If you're behind a restrictive network or proxy, allow that address, or download the file once and host it yourself (updating the install address in package.json). Because it isn't installed from the npm registry, routine dependency-vulnerability scans can't check it automatically — when upgrading, check the SheetJS changelog by hand and update the version in the install address. The lockfile records a checksum of the exact file used, so installs stay consistent.
Deploying: Vercel builds and hosts the site directly from this repository. Pushing to main deploys to production; pull requests get their own preview link. GitHub Actions is the quality gate; Vercel is the host.
Set NEXT_PUBLIC_SITE_URL in the Vercel project once a custom domain is attached — it feeds the sitemap, robots file, and page previews (lib/site.ts).
app/ Home page, tool pages, blog, privacy, terms
components/ Header (search + voice), upload, download, shared tool UI
lib/tools.ts Tool registry (titles, descriptions, how-to steps)
lib/pdf/ File validation, size limits, load/render helpers
lib/p2p/ Connection handling for P2P Share & Whiteboard
lib/workers/ Background worker and the per-tool engines it loads
content/blog/ Written guides
scripts/ Build helpers, screenshot capture, tests, link checking
.github/workflows/ CI configuration
docs/ Documentation assets (screenshot.png)
Files are read into memory in your browser, processed there, and offered back to you as a download. They are never uploaded to OfflinePDF's servers. Passwords used to encrypt or unlock a file never leave your device.
- No upload, by default: 40 of the 42 tools make zero network requests while processing your file.
- P2P Share & Whiteboard, disclosed: these two tools connect two browsers directly for a live session. To set up that connection, the browser contacts a connection-brokering service, which sees IP addresses and session IDs but never file contents or drawings. Once connected, everything passes directly between the two browsers and is never stored on a server.
- Error reporting (optional, set per deployment): if enabled by the site operator, unexpected tool failures send a short, automatically-scrubbed error report — never file data — to help diagnose the issue. This is limited to a small number of reports per visit.
- Camera access: Scan to PDF uses your device camera directly in the page, with no upload.
See /privacy and /terms on the live site for full details.