SDK

Q: Can I edit a document with the SDK?

Not through @paged-media/idml-viewer. It is read-only by design. Editing flows through the editor and the scripting layer.

The Paged SDK is the WebGPU rendering API — load an IDML document, render its pages, and drive a pan/zoom camera — published as @paged-media/idml-viewer. It is the read-only sibling of the editor.

The SDK renders; the editor edits. The Paged SDK is the public, WebGPU-only rendering surface of the engine — it loads an .idml document, lays out its pages, and renders them to a canvas (or to raw pixels), with a camera you drive for pan and zoom. It does not mutate documents or run scripts; that is the editor's job.

In short the SDK ships as @paged-media/idml-viewer — a small TypeScript wrapper over the engine's paged-sdk WebGPU session. You call createViewer({ canvas }), await viewer.load(bytes), and then drive zoom/scroll/goToPage/fit while the engine renders. The same package powers the live preview embedded throughout this documentation, and the standalone community viewer. Because it is renderer-core only — no paged-mutate, no paged-canvas, no paged-script — it is small, embeddable, and safe to drop into any page.

The shape of the API

import { createViewer } from '@paged-media/idml-viewer';

const viewer = await createViewer({ canvas });
await viewer.load(idmlBytes);          // ArrayBuffer | Uint8Array | Blob | URL

viewer.fit('width');                   // fit the page to the viewport
viewer.goToPage(2);                    // jump to a page
viewer.setZoom(1.5, { anchor });       // zoom about a point

viewer.on('pageChanged', ({ page }) => updatePager(page));

What lives where

Layer	Package	Role
TS wrapper	`@paged-media/idml-viewer`	the ergonomic API: `createViewer`, the `Viewer` interface, the camera helpers. The surface you import.
WebGPU session	`@paged-media/sdk` (`ViewerSession`)	the wasm boundary: load, set page, present a camera, render to a canvas or to bytes.
Introspection	`@paged-media/introspect-wasm` (`Inspector`)	read-only document inspection: scene tree, property descriptors, single-page PNG render.

What the SDK does — and does not — do

Does: load IDML, report load diagnostics, lay out pages, render to a <canvas>/OffscreenCanvas or to RGBA bytes at a chosen DPI, drive a camera (zoom, scroll, fit, page navigation, layout mode), and emit events.
Does not: change the document, place or restyle content, run paged.* scripts, or host plugins. For any of those you want the scripting layer or the editor.

Requirements

The SDK renders with WebGPU only — there is no CPU fallback in the shipped viewer. A browser without WebGPU gets a clear "WebGPU unavailable" signal rather than a blank canvas. Embedding the read-only viewer needs nothing special; embedding it alongside the editor's SharedArrayBuffer path additionally needs cross-origin isolation (COOP/COEP) — see the embedding guide.

Read on

Embedding guideinstall, mount a canvas, load a document, meet the requirements.
Camera & navigationzoom, fit, page navigation, layout mode — with a live playground.
Events & errorssubscribe with viewer.on(…) and branch on typed error codes.
Rendering to pixelsthumbnails and headless rasters.
ViewerSessionthe low-level wasm session.
API referenceevery member, with a live filter.

Frequently asked questions

Is the SDK the same thing as the editor? No. The SDK is renderer-core only: it shows a document. The editor adds mutation, scripting, panels, and plugins on top of the same engine.

Can I edit a document with the SDK? Not through @paged-media/idml-viewer. It is read-only by design. Editing flows through the editor and the scripting layer.

Does it work without WebGPU? The shipped viewer is WebGPU-only and reports cleanly when WebGPU is absent. The engine retains a CPU path for headless fidelity testing, but it is not part of the public viewer.