Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.
JavaScript 3d-generationai-agentsanimationavatardigital-humangenerative-aiglbgltfimage-to-3dllmmcpmcp-servermodel-context-protocolsolanatext-to-3dthreejsweb-componentswebglwebxrx402
// /walkthroughs: the index of interactive walkthroughs.
//
// Reads the same generated manifest the player reads, so a walkthrough that
// captured cleanly appears here and one that did not cannot. Cards carry the
// real first frame of each walkthrough as their cover, which means the index
// is also the fastest visual check that the media is current.
const MANIFEST_URL = '/walkthroughs/manifest.json';
function esc(s) {
return String(s ?? '').replace(/[&<>"']/g, (c) => ({ '&': '&', '<': '<', '>': '>', '"': '"', "'": ''' })[c]);
}
function cardHtml(w) {
const cover = w.cover || (w.steps[0] && w.steps[0].shot);
return `
<li class="wt-card" style="--wt-card-accent:${esc(w.accent)}">
<div class="wt-card-shot">
${cover ? `<img src="${esc(cover)}" alt="" loading="lazy" decoding="async" width="1440" height="900" />` : ''}
<span class="wt-card-steps">${w.steps.length} steps</span>
</div>
<div class="wt-card-body">
<h2 class="wt-card-title"><a href="/walkthroughs/${esc(w.slug)}">${esc(w.title)}</a></h2>
<p class="wt-card-outcome">${esc(w.outcome)}</p>
<p class="wt-card-blurb">${esc(w.blurb)}</p>
<ul class="wt-card-meta">
<li class="wt-chip wt-chip-level" style="--wt-accent:${esc(w.accent)}">${esc(w.level)}</li>
<li class="wt-chip">${w.minutes} min</li>
</ul>
</div>
</li>`;
}
function renderMessage(root, heading, detail, retry) {
root.innerHTML = `
<div class="wt-msg">
<h1>${esc(heading)}</h1>
<p>${esc(detail)}</p>
<div class="wt-msg-actions">
<a class="wt-msg-btn" href="/tutorials">Read the tutorials</a>
${retry ? '<button type="button" class="wt-msg-btn wt-msg-btn-ghost" id="wt-idx-retry">Try again</button>' : ''}
</div>
</div>`;
const btn = root.querySelector('#wt-idx-retry');
if (btn) btn.addEventListener('click', boot);
}
async function boot() {
const root = document.getElementById('wt-index');
if (!root) return;
root.innerHTML = `
<ul class="wt-idx-grid" aria-hidden="true">
${'<li class="wt-card"><div class="wt-card-shot wt-sk-stage"></div><div class="wt-card-body"><div class="wt-sk-line wt-sk-line-lg"></div><div class="wt-sk-line wt-sk-line-sm"></div></div></li>'.repeat(4)}
</ul>
<p class="wt-sr" role="status" aria-live="polite">Loading walkthroughs</p>`;
let manifest;
try {
const res = await fetch(MANIFEST_URL, { cache: 'no-cache' });
if (!res.ok) throw new Error(`manifest ${res.status}`);
manifest = await res.json();
} catch (err) {
renderMessage(root, 'Could not load the walkthroughs', `The walkthrough index did not load (${err.message}). The written tutorials cover the same ground.`, true);
return;
}
const list = manifest.walkthroughs || [];
if (!list.length) {
renderMessage(root, 'No walkthroughs published yet', 'Nothing has been captured into the walkthrough library. The written tutorials cover the same ground in the meantime.', false);
return;
}
const captured = manifest.generatedAt ? new Date(manifest.generatedAt) : null;
root.innerHTML = `
<ul class="wt-idx-grid">${list.map(cardHtml).join('')}</ul>
<p class="wt-idx-note">
<strong>These frames are the real product.</strong>
Every picture here was captured by a browser driving <a href="https://three.ws">three.ws</a> itself, and each step is pinned to a live element rather than a hand-drawn box, so a walkthrough breaks loudly when the interface moves instead of quietly going out of date.
${captured ? `Last captured ${esc(captured.toISOString().slice(0, 10))}.` : ''}
</p>`;
}
if (document.readyState === 'loading') document.addEventListener('DOMContentLoaded', boot);
else boot();