Aura3D: The AI-Native 3D SDK for the Web

AuraOne Open now includes Aura3D, an MIT-licensed TypeScript SDK for browser-native 3D scenes, prompt-to-code workflows, typed GLB/glTF assets, WebGL/WebGPU, diagnostics, screenshot evidence, and static deploy checks.

The short version: Aura3D lets a developer or coding agent describe a 3D surface and keep the result as editable TypeScript.

That matters because browser 3D is moving into ordinary software. Product pages need real GLB viewers. Data apps need spatial views. Robotics and Spatial 3D teams need public explainers that do not expose private captures. Agent-generated interfaces need an SDK that gives them maintained scene systems instead of a blank renderer and improvised glue.

Aura3D is built for that workflow.

What Aura3D Gives Developers

Aura3D starts from the way AI coding agents actually work. A prompt should not produce a fragile pile of geometry and hand-written asset paths. It should start from maintained scene kits, use typed assets, and leave behind source code a developer can inspect.

The core install path is simple:

npx create-aura3d@latest my-scene --template product-viewer
cd my-scene
npm run dev

Teams can also install the engine directly:

npm install @aura3d/engine

The SDK includes scene kits for product viewers, particles, material labs, data worlds, city blocks, physics scenes, mini-games, and cinematic browser scenes. It also gives agents clear rules for typed assets, route health, screenshot checks, and static deploy output.

Typed Assets, Not Invented Paths

3D workflows break when a model invents an asset ID or points at a model that was never supplied. Aura3D's safe asset workflow makes the supplied GLB or glTF explicit before the app code uses it.

import { createAuraApp, sceneKits } from "@aura3d/engine";
import { assets } from "./aura-assets";

const kit = sceneKits.productViewer(assets.product);

createAuraApp("#app", kit.toAppOptions());

That pattern is small, but it is important. The generated source owns the scene. The asset reference is inspectable. The deployed app remains normal browser software.

How It Fits AuraOne Open

AuraOne Open is the public tooling layer behind Human Data OS and App Data OS. Rubric Studio Open handles rubric authoring. Agent Studio Open handles agent traces and regressions. Robotics Studio Open handles local robotics dataset review. Trust Toolkit packages the review and reliability checks. AuraGlass gives the UI layer.

Aura3D adds the browser-native 3D layer.

It belongs in the Open catalog because 3D product surfaces, data worlds, robotics explainers, and Spatial 3D previews need the same open-source contract as the rest of AuraOne Open: inspect the source, run it locally, keep the artifact, and bring AuraOne in only when shared state, review, or governance is the real problem.

Aura3D vs Spatial 3D Lab

The boundary is simple.

Aura3D is the open SDK for building the scene: a product viewer, data world, interactive explainer, or browser-native 3D app.

Spatial 3D Lab is the governed App Data workflow for real 3D work: capture lineage, transform history, alignment review, delivery approvals, reviewer sign-off, and tuned weights.

One helps developers ship browser 3D. The other helps teams prove what happened to private 3D captures and delivery packets. They are connected, but they should not be confused.

Start Here

Launch the site at aura3d.auraone.ai, open the AuraOne-owned overview at /open/aura3d, or inspect the source at github.com/auraoneai/aura3d.

If you are using an AI coding assistant, start with the agent context:

https://aura3d.auraone.ai/llms.txt

The goal is not to hide 3D inside a black box. The goal is to make browser 3D feel like software again: prompted, shaped, inspected, tested, and shipped as source.