System design

A static application with local document workers

Node.js builds the application; it does not process customer documents. The production output is ordinary HTML, CSS, JavaScript, WebAssembly, and on-demand OCR data. Optional AI add-ons are separate deployments.

Static Next.js shell

React, TypeScript, Tailwind, static export, PWA service worker, and no server actions or API routes.

PDF engine adapter

A working pdf-lib engine ships by default. Apryse and other commercial engines plug into a documented boundary.

Operation graph

Validated, serializable operations power quick tools, Studio, history, undo, redo, and future workflow automation.

Browser storage

OPFS stores revision snapshots; IndexedDB stores operation metadata, preferences, and recent-session information.

OCR and scanning

Tesseract.js OCR and lightweight scan cleanup run locally in dedicated workers with on-demand English language data.

Optional local AI

Local AI is not bundled in the core editor; audited add-ons can be deployed separately when required.

Cryptography worker

WebCrypto performs hashing and local signatures; PKI.js parses X.509 structures behind a worker boundary.

Zero-document-network tests

Integration browser checks exercise local document operations, restricted ad traffic, and rejection of unrelated cross-origin requests.

Capability boundaries

The runnable open-source build is intentionally honest about the functions that require a professional PDF SDK or a separately audited add-on.

Capability	Included engine	Adapter path
Page organisation, watermark, numbering, metadata	Working locally	Optional commercial enhancement
PDF rendering and page images	PDF.js worker	Apryse renderer optional
True existing-text reflow editing	Not claimed	Apryse adapter
Certificate-backed PAdES signing	Crypto foundation only	Apryse or dedicated signing SDK + audit
Office conversion	Adapter scaffold	Self-hosted ZetaOffice or licensed SDK
Local AI	Not bundled	Deploy a separate audited add-on