Open-Source Virtual Tour Landing Page Seijaku and Eight Three.js Agent Skills

X postMeng ToResource

Meng To has open-sourced Seijaku, an interactive 3D virtual tour landing page of a Japanese house. Alongside the project, he released eight modular Three.js skills designed to steer AI coding agents when generating complex 3D scenes.

Key points

  1. The Seijaku landing page is built with Three.js using Fable 5.1 and the dream-loop skill.
  2. According to Meng To, the code base is under 1 MB excluding assets, featuring scroll-driven room tours, seasonal changes, realistic water with koi, and light rays.
  3. Eight dedicated Three.js skills cover room-to-room scroll tours, sky rays, sky backgrounds, falling leaves, four seasons, high-resolution textures, high-poly models, and retina rendering.
  4. The underlying dream-loop technique prompts an agent to create a target image, compare the live 3D output against it using a critic, and iteratively refine the code.

Why it matters

Generating 3D web scenes with LLMs typically requires heavy steering. Packaging rendering techniques into modular agent skills helps automated workflows reliably implement Three.js features without manual trial and error.

Analysis

Meng To demonstrates a practical pattern for agent-assisted 3D development. Instead of publishing code alone, providing concrete agent skills addresses the steep prompt-engineering barrier in WebGL. The combination of image-guided evaluation loops and explicit capability prompts shows how developers can structure complex visual coding tasks for AI agents.

Original post

Meng To

@MengTo · 16 September 2026

I open-sourced this scrolling virtual-tour landing page, along with 8 new three.js skills. They cover room-to-room tours controlled by scrolling, sky rays, sky backgrounds, falling leaves, four seasons, high-resolution textures, high-poly models, and retina rendering. Live t.co/wkwwFcig4s t.co/P4BVZmrYWG

275 likes12 replies260 bookmarks21.4k views

View on X
More posts in the thread (6)
  1. @eamin_ux 3d needs a ton of steering right now. so these skills should help.
  2. @Duy_ZzZz you can try on t.co/pzTjpCw8XH works quite well
  3. @islemonarch haha yea, everyone's running out of tokens these days
  4. @MujovicEldin have you tried it? works fine on my phone. you can improve perf with a prompt.
  5. @Navlio that's easily solved with html code.
  6. @1919Nooah i underestimate how useful the skills are for my agents. sometimes i just forget about them and my agents pick them up.
Selected replies (5)
  • @MujovicEldin @MengTo It’s not sustainable, it doesn’t work on most of devices. It’s slow and eats a lot of memory. Anyway it looks really good.
  • @Navlio @MengTo worth a line in the readme for anyone shipping this as an actual landing page: it's all canvas, so a crawler and an assistant both arrive at a page with no words in it the pitch still needs to exist as real text somewhere down the page
  • @1919Nooah @MengTo Packaging it as skills instead of just a repo is the part people will actually use. Most open source drops leave you to reverse-engineer the "how", you shipped the "how" directly.
  • @eamin_ux @MengTo Visual detail needs clear purpose
  • @islemonarch @MengTo Oh i’m waiting for my reset to give this a try, thanks for sharing Meng.

Mentioned links

Links and tools from this entry

Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • Repository:achimala/dream-loop

    dream-loop: Agent Skill for 3D Visuals via Iterative Critic Feedback

    dream-loop is an open-source agent skill by Anshu Sharma (@anshuc) designed to build high-quality 3D scenes, games, and web apps through a closed feedback loop. Image generation creates a reference target, after which an AI critic visually inspects live output and directs iterative refinements.

    1.5kstarsJavaScript

    AI & LLMs· Tool

  • Repository:MengTo/Skills

    Agent Skills: Reusable Playbooks for AI Coding Agents

    The GitHub repository MengTo/Skills provides a curated collection of Markdown-based 'agent skills'. It enables designers and builders to guide AI coding agents like Codex, Claude, and Cursor through complex web design, interaction, and 3D rendering tasks.

    6.2kstarsHTML

    AI & LLMs· Resource

  • Repository:MengTo/seijaku

    Seijaku: Interactive Three.js 3D Walkthrough of a Kyoto Residence

    Seijaku is an interactive web demo by Meng To that delivers a continuous, scroll-based walkthrough of a virtual Kyoto residence. It combines procedural Three.js architecture, Web Audio soundscapes, and interactive material studies without requiring a build step.

    32starsHTML

    Web development· Demo

  • X post:Meng To

    Anatomy: Agent Skill for Animated Isometric SVG Explainers

    Meng To showcases 'Anatomy', an open-source agent skill by developer Ryan (@wheresryan22) that enables coding agents like Claude Code to generate interactive isometric SVG and WebGL visualisations.

    3.4kbookmarks101.5kviews

    AI & LLMs· Tool

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.