Use Aholo Viewer logo

Use Aholo Viewer

OrganizationPopular
manycoretech
use-aholo-viewer

Guide external npm package users and AI coding agents integrating @manycore/aholo-viewer into web applications. Use when asked to install, bootstrap, render 3D Gaussian Splatting or mesh content, configure cameras, scenes, lighting, viewer pipeline options, or write TypeScript examples for the public @manycore/aholo-viewer npm package.

Overview

Publishermanycoretech
Repositoryaholo-viewer
Skill nameuse-aholo-viewer
Stars
1.1K
Forks
113
Bundled files
2
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 2 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by manycoretech on GitHub. Read the source before you install it.

Installation

Install the Use Aholo Viewer AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/manycoretech/aholo-viewer/tree/master/docs/ai/skills/use-aholo-viewer
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/manycoretech/aholo-viewer.git /tmp/aholo-viewer
mkdir -p .claude/skills
cp -r /tmp/aholo-viewer/docs/ai/skills/use-aholo-viewer .claude/skills/use-aholo-viewer
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Use Aholo Viewer in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Use Aholo Viewer on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Use Aholo Viewer is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Use Aholo Viewer

Use this skill to build or modify user applications that consume the public @manycore/aholo-viewer npm package. Treat it as a full-featured browser rendering package, not as this repository's internal renderer source.

Source Docs

Before generating non-trivial code, read the AI-oriented docs advertised by https://aholojs.dev/llms.txt:

  • https://aholojs.dev/llm/manual/getting-started.md for installation and the minimal Vite setup.
  • https://aholojs.dev/llm/manual/basic-concepts.md for scenes, objects, cameras, viewports, and the render flow.
  • https://aholojs.dev/llm/api/index.md to confirm public exports.
  • https://aholojs.dev/llm/examples/index.md for runnable TypeScript patterns.

Prefer these Markdown docs over scraping rendered HTML pages.

Integration Workflow

  1. Install the package in a browser app:

    bash
    npm install --save @manycore/aholo-viewer
    npm install --save-dev vite typescript
  2. Import only public package exports from @manycore/aholo-viewer.

  3. Create a real DOM container with stable dimensions before calling createViewer. Attach application DOM event listeners to this container, not to the engine canvas.

  4. Create a camera, position it, call lookAt, add objects to viewer.getScene(), then call viewer.setCamera(camera).

  5. Configure rendering with setViewerConfig(viewer, { pipeline: ... }).

  6. Render explicitly with viewer.render(). If the app relies on viewer-triggered invalidation, set viewer.requestRenderHandler to schedule another render with requestAnimationFrame.

  7. Clean up when removing content: detach scene objects, stop render/update loops that still reference them, then release GPU resources or owned resources according to the cleanup rules below.

Minimal Splat Example

Use this pattern for a Vite/TypeScript browser entry:

ts
import {
    BackgroundMode,
    Color,
    PerspectiveCamera,
    SplatLoader,
    SplatUtils,
    Vector3,
    createViewer,
    setViewerConfig,
} from '@manycore/aholo-viewer';

const SPLAT_URL = 'https://holo-cos.aholo3d.cn/aholo-opensource/gs_file/bear/bear.3d71a266.sog';

const container = document.createElement('div');
container.style.width = '500px';
container.style.height = '500px';
document.body.appendChild(container);

async function main() {
    const viewer = createViewer('example-viewer', container, {});
    const camera = new PerspectiveCamera(60, 1, 0.1, 2000);

    const response = await fetch(SPLAT_URL);
    const buffer = await response.arrayBuffer();
    const data = await SplatLoader.parseSplatData(
        SplatLoader.SplatFileType.SOG,
        new Uint8Array(buffer),
        SplatLoader.SplatPackType.Compressed,
    );
    const splat = await SplatUtils.createSplat(data);

    camera.up.set(0, -1, 0);
    camera.position.set(-1.5, -0.5, 0);
    camera.lookAt(new Vector3(0, 0, 0));

    viewer.getScene().add(splat);
    viewer.setCamera(camera);
    setViewerConfig(viewer, {
        pipeline: {
            Background: {
                background: {
                    active: BackgroundMode.BasicBackground,
                    basic: { color: new Color(0, 0, 0) },
                },
                ground: { enabled: false },
            },
            Splatting: { enabled: true },
            TAA: { enabled: false },
        },
    });

    const render = () => viewer.render();
    viewer.requestRenderHandler = () => requestAnimationFrame(render);
    requestAnimationFrame(render);
}

main();

Common Public Exports

  • Viewer setup: createViewer, setViewerConfig, Viewer, Viewport.
  • Cameras and math: PerspectiveCamera, OrthographicCamera, Vector2, Vector3, Vector4, Color, Matrix4, Quaternion.
  • Scene objects: Object3D, Scene3D, Mesh, Splat, AmbientLight, DirectionalLight, PointLight, SpotLight.
  • Mesh content: BufferGeometry, BufferAttribute, MeshBasicMaterial, MeshPhongMaterial, Side, downloadTexture.
  • Splat content: SplatLoader, SplatUtils, CompressedSplat, SuperCompressedSplat, SogSplat.
  • Optional loaders/utilities: GLTFLoader, DracoLoader, Animation, Events.

Confirm exact symbols in /llm/api/index.md when using less common APIs.

Patterns

For 3DGS content:

  • Use SplatLoader.parseSplatData(...) to parse supported splat formats.
  • Use SplatUtils.createSplat(data) to create a renderable splat.
  • Enable pipeline.Splatting.
  • Set camera up vectors deliberately. OpenCV-style 3DGS data commonly uses camera.up.set(0, -1, 0).

For mesh content:

  • Build geometry with BufferGeometry, BufferAttribute, and optional indices.
  • Use MeshPhongMaterial with lights for shaded meshes, or MeshBasicMaterial for unlit objects.
  • Add lights such as AmbientLight, DirectionalLight, or PointLight before rendering lit materials.

For interactive apps:

  • Tie camera controls or animations to a frame loop that calls viewer.render().
  • Keep viewer container size stable and update rendering after layout changes.
  • Register pointer, wheel, keyboard focus, drag/drop, resize, and custom DOM event listeners on the viewer container or another application-owned element.
  • Cache large splat data when appropriate, but keep the first example simple unless the user asks for caching.

DOM Events and Canvas Access

Treat the container as the stable application boundary:

  • Listen for DOM events on the container passed to createViewer(...), or on another element owned by the application.
  • Do not attach event listeners to the viewer canvas. The canvas is managed internally by the engine and may be replaced.
  • Do not persistently store the engine canvas element in application state, class fields, module globals, framework refs, or closures intended to outlive the current operation.
  • If direct canvas access is unavoidable, read it just-in-time from the current viewer/container state, use it immediately, and discard the reference.
  • Remove container-level event listeners during teardown before releasing viewer or scene resources.

Resource Cleanup

Use the lightest cleanup method that matches ownership and lifecycle:

  • Use freeGPU() when the goal is to release GPU memory while keeping the object instance logically reusable. The object may upload GPU resources again if it is rendered or updated later.
  • Use freeAllGpuResourceOwned() when the object owns related GPU resources that should also be released. For example, geometry.freeAllGpuResourceOwned() releases the geometry GPU resource plus its attribute and index GPU resources; drawable.freeAllGpuResourceOwned() releases its materials, geometry, and drawable GPU resources.
  • Use destroy() only when the object is permanently retired. Before calling destroy(), ensure the object is removed from the scene, no frame callback, async load, event listener, control, viewport, material, geometry, or app state will use it again, and it is not currently participating in a render/update pass.
  • Prefer freeGPU() or freeAllGpuResourceOwned() for memory-pressure cleanup. Prefer destroy() for final teardown after references are severed.
  • In component frameworks, cancel pending async work and stop animation/requestRender callbacks before releasing resources in unmount cleanup.

Guardrails

  • Do not hand-write code against private renderer internals or repository-only paths.
  • Do not assume a server-side runtime; the viewer needs browser DOM and WebGL/WebGL2-capable canvas support.
  • Do not construct Splat directly. Use loader and utility APIs.
  • Do not register app DOM events on the engine canvas. Register on the container because the engine owns the canvas lifecycle and may replace it.
  • Do not hold persistent references to the engine canvas. Read it only when needed, use it immediately, and release the reference.
  • Do not call destroy() on resources that shared objects, cached data, render callbacks, or pending async work may still reference.
  • Do not use destroy() as the default GPU-memory cleanup method. Use freeGPU() for the object's own GPU state, or freeAllGpuResourceOwned() when owned child resources should also release GPU state.
  • Do not omit cleanup in component frameworks. In React, Vue, Svelte, or similar frameworks, dispose viewer-owned objects and cancel pending async work during unmount.
  • Do not invent package exports. Check the LLM API index or installed package types.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Use Aholo Viewer AI skill do?

Guide external npm package users and AI coding agents integrating @manycore/aholo-viewer into web applications. Use when asked to install, bootstrap, render 3D Gaussian Splatting or mesh content, configure cameras, scenes, lighting, viewer pipeline options, or write TypeScript examples for the public @manycore/aholo-viewer npm package.

Why use Use Aholo Viewer on TypingMind?

Because you install it once and use it with any model. Use Aholo Viewer is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Use Aholo Viewer in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/manycoretech/aholo-viewer/tree/master/docs/ai/skills/use-aholo-viewer. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Use Aholo Viewer?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Use Aholo Viewer?

As many as you like. As long as a model supports skills, you can use Use Aholo Viewer with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Use Aholo Viewer AI skill free?

Yes. It is published on GitHub by manycoretech under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇