Skip to content

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.

docs(figures): show the page each tutorial is about instead of only claiming a figure exists

nirholas303c17ce3aa6
diff --git a/docs/sign-language.md b/docs/sign-language.md
index 94e68c811..9879aa2cd 100644
--- a/docs/sign-language.md
+++ b/docs/sign-language.md
@@ -4,6 +4,10 @@ three.ws avatars communicate in American Sign Language. Words that have a sign a
New here? The step-by-step walkthrough is [Tutorial: make your avatar sign](https://three.ws/tutorials/sign-with-your-avatar). This page is the reference: every surface, the developer API, and how the engine works.
+[![A rigged 3D avatar mid-sign on the three.ws sign-language page, beside the phrase input and the speed and signing-hand controls](/docs/img/sign-language-hero.webp)](/sign-language)
+
+*Type anything: the avatar signs the words it has signs for and spells the rest.*
+
## Start here
| I want to | Go here | What happens |
@@ -102,6 +106,10 @@ Another 41 everyday spellings route to those same signs, so ordinary sentences w
[/asl-alphabet](https://three.ws/asl-alphabet) is the reference for the manual alphabet itself. Every letter A-Z and digit 0-9 is formed by a live rig you can orbit, so you see the handshape from any angle rather than from one photographed side.
+[![A rigged 3D avatar forming a letter of the American manual alphabet, beside that letter's description and the speed and signing-hand controls](/docs/img/asl-alphabet-hero.webp)](/asl-alphabet)
+
+*Every letter of the manual alphabet on a hand you can turn around.*
+
- **Click a key, or press it.** Typing `q` signs a Q. A single letter holds its pose so you can study it; a word settles back to rest the way a signer finishes.
- **Every letter is described**, with the letters it is confused with named explicitly (F against 9, M against N, K against V, G against Q). Look-alikes are where reading breaks down, so they are called out rather than left to be discovered.
- **Spell any word** and each key lights up as the hand reaches it. The word is also laid out under the input, one cell per letter, lit in the same cadence, so a long word can be followed without scrolling back to the A-to-Z grid. The highlight comes from the clip builder itself, not from a timer guessed alongside it, so it stays in step at any speed.
diff --git a/docs/tutorials/animate-your-agent.md b/docs/tutorials/animate-your-agent.md
index 29b0a4056..4c4b99597 100644
--- a/docs/tutorials/animate-your-agent.md
+++ b/docs/tutorials/animate-your-agent.md
@@ -6,6 +6,10 @@ By the end of this tutorial your agent will move on its own: nod when a model fi
**Time:** about 12 minutes.
+[![The Animation Studio timeline with an avatar mid-clip](/docs/img/animation-studio.webp)](/pose)
+
+*Animation Studio drives the same canonical clip library the embeds use.*
+
---
## What you're building
diff --git a/docs/tutorials/animate-your-avatar.md b/docs/tutorials/animate-your-avatar.md
index 87eba4e3f..059373d69 100644
--- a/docs/tutorials/animate-your-avatar.md
+++ b/docs/tutorials/animate-your-avatar.md
@@ -4,6 +4,10 @@ By the end of this tutorial you'll be able to take any rigged avatar, browse the
**Prerequisites:** a three.ws account and at least one rigged avatar ([create one](/create), or load any public avatar). No code required, everything here happens in the browser. Light JavaScript familiarity helps only for the optional "how retargeting works" section.
+[![The Animation Studio timeline with an avatar mid-clip](/docs/img/animation-studio.webp)](/pose)
+
+*Animation Studio drives the same canonical clip library the embeds use.*
+
---
## What you're building
diff --git a/docs/tutorials/build-a-scene.md b/docs/tutorials/build-a-scene.md
index 1b368af5c..a990e7140 100644
--- a/docs/tutorials/build-a-scene.md
+++ b/docs/tutorials/build-a-scene.md
@@ -6,6 +6,10 @@ Along the way you'll learn how the editor's panels fit together, why the scene a
**Prerequisites:** a desktop browser (the editor is keyboard- and gizmo-driven). At least one GLB or glTF model to import — generate one for free from text at [/forge](/forge), or use any GLB file you already have. No account required.
+[![Scene Studio with a 3D scene in the viewport and the object rail](/docs/img/scene-studio.webp)](/scene)
+
+*Scene Studio composes objects and avatars into one shareable scene.*
+
---
## What you're building
diff --git a/docs/tutorials/character-library-to-embed.md b/docs/tutorials/character-library-to-embed.md
index 8d49c6fcc..ee9adf5ef 100644
--- a/docs/tutorials/character-library-to-embed.md
+++ b/docs/tutorials/character-library-to-embed.md
@@ -20,6 +20,10 @@ Ten minutes is not marketing. The list is: two minutes reading the library manif
- No account, no API key, no payment. The library is free and public, and so is the embed runtime.
- Background reading, if you want it: [Character Library](/docs/character-library.md) for the gallery and manifest, [Embedding Guide](/docs/embedding.md) for the full `<agent-3d>` contract.
+[![The character library grid of ready-to-use rigged characters](/docs/img/character-library.webp)](/character-library)
+
+*Every character here is a real rigged GLB you can embed as-is.*
+
---
## Step 1 - Read the library
diff --git a/docs/tutorials/claim-threews-name.md b/docs/tutorials/claim-threews-name.md
index 9fa12871e..b5ed790fe 100644
--- a/docs/tutorials/claim-threews-name.md
+++ b/docs/tutorials/claim-threews-name.md
@@ -6,6 +6,10 @@ Along the way you'll understand what the subdomain points at on-chain, why the m
**Prerequisites:** a signed-in three.ws account with a **username set** (the subdomain label must match it) and at least one agent — creating an agent provisions the Solana wallet that receives the name ([create one](/create)). Light JavaScript familiarity for the pay-by-name section; the claim itself is no-code.
+[![The threews.sol subdomain claim page with the name availability field](/docs/img/names-claim.webp)](/threews/claim)
+
+*Claiming a name is one field and one signature.*
+
---
## What you're building
diff --git a/docs/tutorials/custom-skill.md b/docs/tutorials/custom-skill.md
index 619a46326..ccbc4358e 100644
--- a/docs/tutorials/custom-skill.md
+++ b/docs/tutorials/custom-skill.md
@@ -6,6 +6,10 @@ Along the way you'll understand the skill bundle format, how the LLM decides whe
**Prerequisites:** JavaScript familiarity, a text editor, and a place to host static files (a free GitHub Pages or Vercel account works fine).
+[![The skills marketplace listing paid and free agent skills](/docs/img/skills-marketplace.webp)](/skills)
+
+*Skills are sold, trialled and installed from this catalogue.*
+
---
## What you're building
diff --git a/docs/tutorials/customize-appearance.md b/docs/tutorials/customize-appearance.md
index 813a71d69..8e72f1861 100644
--- a/docs/tutorials/customize-appearance.md
+++ b/docs/tutorials/customize-appearance.md
@@ -13,6 +13,10 @@ This tutorial walks through every appearance attribute the `<agent-3d>` element
**Prerequisites:** A page with the embed working from [Embed in 30 seconds](/tutorials/embed-in-30-seconds). You can write the attributes here straight into that page.
+[![The three.ws Avatar Studio with a rigged avatar in the viewport and the customisation rail](/docs/img/avatar-studio.webp)](/avatar-studio)
+
+*Avatar Studio: the rig on the left, every appearance control on the right.*
+
---
## Step 1 — How the element reads your attributes
diff --git a/docs/tutorials/embed-in-30-seconds.md b/docs/tutorials/embed-in-30-seconds.md
index 3e5c3e065..a127b6da4 100644
--- a/docs/tutorials/embed-in-30-seconds.md
+++ b/docs/tutorials/embed-in-30-seconds.md
@@ -12,6 +12,10 @@ This tutorial walks through those two lines end to end. By the time you finish,
**Prerequisites:** A text editor and a browser. You do not need Node, npm, a framework, or any 3D experience.
+[![The Widget Studio configurator with a live avatar preview and generated embed code](/docs/img/widget-studio.webp)](/studio)
+
+*Widget Studio writes the embed snippet while you watch the result.*
+
---
## Step 1 — Get an agent ID
diff --git a/docs/tutorials/embed-on-website.md b/docs/tutorials/embed-on-website.md
index a58158158..c04718b84 100644
--- a/docs/tutorials/embed-on-website.md
+++ b/docs/tutorials/embed-on-website.md
@@ -9,6 +9,10 @@ This tutorial walks you through every path for adding a three.ws agent to your s
- Add it to Webflow, WordPress, or Squarespace
- Show or hide the agent based on scroll position, page URL, or a timer
+[![The Widget Studio configurator with a live avatar preview and generated embed code](/docs/img/widget-studio.webp)](/studio)
+
+*Widget Studio writes the embed snippet while you watch the result.*
+
---
## Before you start
diff --git a/docs/tutorials/explore-docs-world.md b/docs/tutorials/explore-docs-world.md
index e5665ce6b..2d8cac198 100644
--- a/docs/tutorials/explore-docs-world.md
+++ b/docs/tutorials/explore-docs-world.md
@@ -2,6 +2,10 @@
Time: about 5 minutes. You need a browser and nothing else. By the end you will have walked the documentation, read a page inside the scene, and entered the world wearing a custom avatar.
+[![Animated loop of the 3D Docs World with the guide avatar on the plaza](/docs/img/docs-world-loop.webp)](/docs/world)
+
+*The documentation as a place you can walk through.*
+
## 1. Enter the world
Open [three.ws/docs/world](https://three.ws/docs/world).
diff --git a/docs/tutorials/first-agent.md b/docs/tutorials/first-agent.md
index 0ad9b269b..58c850153 100644
--- a/docs/tutorials/first-agent.md
+++ b/docs/tutorials/first-agent.md
@@ -10,6 +10,10 @@ By the end of this tutorial you will have a live, talking 3D character on a page
**Prerequisites:** Basic HTML and JavaScript. You don't need to know Three.js, WebGL, or anything 3D-specific.
+[![The three.ws getting-started checklist showing the first setup steps](/docs/img/start-checklist.webp)](/start)
+
+*The getting-started checklist is the shortest path from account to first agent.*
+
---
## Step 1 — Pick a 3D avatar
diff --git a/docs/tutorials/first-prompt-to-3d.md b/docs/tutorials/first-prompt-to-3d.md
index 0c9e2cb92..645c276b6 100644
--- a/docs/tutorials/first-prompt-to-3d.md
+++ b/docs/tutorials/first-prompt-to-3d.md
@@ -11,6 +11,10 @@ Type one sentence into ChatGPT and get a real, textured 3D model back: spin it i
**Prerequisites:** A ChatGPT account. A smartphone (iPhone or Android) for the AR step; every other step works on a computer.
+[![The three.ws prompt-to-3D page with the description field and quality tier picker](/docs/img/forge-prompt-panel.webp)](/create/prompt)
+
+*Describe it to 3D: the prompt field and the tier picker are the whole interface.*
+
---
## Step 1: Add the 3D Studio to ChatGPT
diff --git a/docs/tutorials/generate-3d-api.md b/docs/tutorials/generate-3d-api.md
index 1a9f2ac4e..b2ddbd2ba 100644
--- a/docs/tutorials/generate-3d-api.md
+++ b/docs/tutorials/generate-3d-api.md
@@ -6,6 +6,10 @@ Everything the [Forge](/forge) does in the browser is available as a plain HTTP
**Prerequisites:** comfortable with `fetch` or `curl`. Examples use Node.js 18+ (built-in `fetch`), but any language with an HTTP client works the same way.
+[![The Forge page showing text and image to 3D generation options](/docs/img/forge-hero.webp)](/forge)
+
+*The Forge accepts a sentence or an image and returns a textured GLB.*
+
---
## The flow in one picture
diff --git a/docs/tutorials/getting-started.md b/docs/tutorials/getting-started.md
index 4f73533bd..9ef81439f 100644
--- a/docs/tutorials/getting-started.md
+++ b/docs/tutorials/getting-started.md
@@ -6,6 +6,10 @@ This is the easiest entry point on the platform. Everything you do here is edita
**Prerequisites:** a modern browser (Chrome, Firefox, or Safari) and an email address. No coding experience, no 3D files, and no crypto required. You'll sign in once, partway through, to save your work.
+[![The three.ws getting-started checklist showing the first setup steps](/docs/img/start-checklist.webp)](/start)
+
+*The getting-started checklist is the shortest path from account to first agent.*
+
---
## What you're building
diff --git a/docs/tutorials/greeting-and-first-speech.md b/docs/tutorials/greeting-and-first-speech.md
index 4f019a20e..2525f012a 100644
--- a/docs/tutorials/greeting-and-first-speech.md
+++ b/docs/tutorials/greeting-and-first-speech.md
@@ -15,6 +15,10 @@ This tutorial covers the greeting flow end to end: waiting for the `agent:ready`
**Prerequisites:** A page with the embed working from [Embed in 30 seconds](/tutorials/embed-in-30-seconds). Your agent should already be saved with a brain configured.
+[![The Voice Lab page with the voice catalogue and preview controls](/docs/img/voice-lab.webp)](/voice)
+
+*Voice Lab previews every voice the speech and lipsync stack can drive.*
+
---
## Step 1 — The minimal greeting
diff --git a/docs/tutorials/image-to-3d.md b/docs/tutorials/image-to-3d.md
index 84ada01cb..31ce7eec5 100644
--- a/docs/tutorials/image-to-3d.md
+++ b/docs/tutorials/image-to-3d.md
@@ -6,6 +6,10 @@ Have a real object you want in 3D? Photograph it from a few angles, drop the pho
**Prerequisites:** none. A phone camera is plenty.
+[![The image-to-3D page with its upload dropzone](/docs/img/image-to-3d-page.webp)](/image-to-3d)
+
+*Drop a photo here and the same pipeline returns a mesh instead of a render.*
+
---
## Step 1 — Take good photos
diff --git a/docs/tutorials/import-avatar-url.md b/docs/tutorials/import-avatar-url.md
index 984343ce2..6a9e08243 100644
--- a/docs/tutorials/import-avatar-url.md
+++ b/docs/tutorials/import-avatar-url.md
@@ -6,6 +6,10 @@ This tutorial walks the URL-import path end to end. It's the sibling of [Upload
**Prerequisites:** a three.ws account (the importer redirects you to sign in if you're not), and a live, publicly reachable URL to a glTF 2.0 `.glb` file. No command line, no Blender, no local tooling required — the whole flow runs in the browser.
+[![The public avatar gallery grid](/docs/img/gallery-grid.webp)](/gallery)
+
+*Every avatar in the gallery is a real GLB with a public render URL.*
+
---
## What you're building
diff --git a/docs/tutorials/prompts-for-3d.md b/docs/tutorials/prompts-for-3d.md
index d6f74dfd0..eff39291a 100644
--- a/docs/tutorials/prompts-for-3d.md
+++ b/docs/tutorials/prompts-for-3d.md
@@ -16,6 +16,10 @@ a sci-fi combat helmet, brushed metal
If you haven't generated your first model yet, do the five-minute [text-to-3D tutorial](/tutorials/text-to-3d) first, then come back here for ideas.
+[![The Forge page showing text and image to 3D generation options](/docs/img/forge-hero.webp)](/forge)
+
+*The Forge accepts a sentence or an image and returns a textured GLB.*
+
---
## The four rules
diff --git a/docs/tutorials/sell-a-skill-with-a-trial.md b/docs/tutorials/sell-a-skill-with-a-trial.md
index 6572169a9..bf0cb54f5 100644
--- a/docs/tutorials/sell-a-skill-with-a-trial.md
+++ b/docs/tutorials/sell-a-skill-with-a-trial.md
@@ -8,6 +8,10 @@ The interesting part is not the listing. It is the middle. A free trial that nev
**Time:** about 15 minutes.
+[![The skills marketplace listing paid and free agent skills](/docs/img/skills-marketplace.webp)](/skills)
+
+*Skills are sold, trialled and installed from this catalogue.*
+
---
## What you're building
diff --git a/docs/tutorials/share-your-agent.md b/docs/tutorials/share-your-agent.md
index ef3eb4eb5..f51194b19 100644
--- a/docs/tutorials/share-your-agent.md
+++ b/docs/tutorials/share-your-agent.md
@@ -13,6 +13,10 @@ This tutorial covers the full sharing surface: finding the public URL, generatin
**Prerequisites:** A saved agent on three.ws. The agent should have a body and a name — covered in [Embed in 30 seconds](/tutorials/embed-in-30-seconds) and [Pick and swap an avatar in Studio](/tutorials/swap-avatar-in-studio).
+[![The three.ws marketplace grid of published agents](/docs/img/marketplace-grid.webp)](/marketplace)
+
+*Published agents land here, forkable by anyone.*
+
---
## Step 1 — Find your agent's public URL
diff --git a/docs/tutorials/swap-avatar-in-studio.md b/docs/tutorials/swap-avatar-in-studio.md
index c2a62c898..3d848a798 100644
--- a/docs/tutorials/swap-avatar-in-studio.md
+++ b/docs/tutorials/swap-avatar-in-studio.md
@@ -12,6 +12,10 @@ This tutorial covers both halves of that. You swap the body in the agent editor'
**Prerequisites:** A saved agent on three.ws. If you've worked through [Embed in 30 seconds](/tutorials/embed-in-30-seconds), you already have one.
+[![The three.ws Avatar Studio with a rigged avatar in the viewport and the customisation rail](/docs/img/avatar-studio.webp)](/avatar-studio)
+
+*Avatar Studio: the rig on the left, every appearance control on the right.*
+
---
## Step 1: Know which surface does what
diff --git a/docs/tutorials/text-to-3d.md b/docs/tutorials/text-to-3d.md
index 00cf567f5..4c1121e9e 100644
--- a/docs/tutorials/text-to-3d.md
+++ b/docs/tutorials/text-to-3d.md
@@ -6,6 +6,10 @@ Type a sentence, get a downloadable 3D model. That's the whole tutorial. By the
**Prerequisites:** none. No account, no wallet, no code. Just a browser.
+[![The three.ws prompt-to-3D page with the description field and quality tier picker](/docs/img/forge-prompt-panel.webp)](/create/prompt)
+
+*Describe it to 3D: the prompt field and the tier picker are the whole interface.*
+
---
## Step 1 — Open the Forge
diff --git a/docs/tutorials/upload-custom-glb.md b/docs/tutorials/upload-custom-glb.md
index efa0a7cdc..43bc6a653 100644
--- a/docs/tutorials/upload-custom-glb.md
+++ b/docs/tutorials/upload-custom-glb.md
@@ -12,6 +12,10 @@ This tutorial covers the whole pipeline. Where compatible GLBs come from, what m
**Prerequisites:** Comfort with a file manager and the command line. Some familiarity with Blender helps for the optional baking step but is not required. You should have an existing agent in [three.ws/my-agents](https://three.ws/my-agents) to swap a body into.
+[![The public avatar gallery grid](/docs/img/gallery-grid.webp)](/gallery)
+
+*Every avatar in the gallery is a real GLB with a public render URL.*
+
---
## Step 1: Understand what the runtime expects
diff --git a/docs/tutorials/view-in-ar.md b/docs/tutorials/view-in-ar.md
index e32ff6c27..b7f460de4 100644
--- a/docs/tutorials/view-in-ar.md
+++ b/docs/tutorials/view-in-ar.md
@@ -11,6 +11,10 @@ This tutorial walks you through every step of seeing a three.ws model in augment
**Prerequisites:** A smartphone (iPhone or Android). For the embedding section, a basic HTML file you can edit.
+[![The Walk experience on a phone-sized viewport with the avatar on screen](/docs/img/walk-mobile.webp)](/walk)
+
+*Walk on a phone: the same rig, the same clips, thumb-sized controls.*
+
---
## Step 1 — Get a 3D model
diff --git a/docs/tutorials/voice-and-lipsync.md b/docs/tutorials/voice-and-lipsync.md
index c278e2510..bdb196f8c 100644
--- a/docs/tutorials/voice-and-lipsync.md
+++ b/docs/tutorials/voice-and-lipsync.md
@@ -6,6 +6,10 @@ Along the way you'll understand why three.ws extracts visemes from the audio its
**Prerequisites:** a three.ws account with at least one agent ([create one](/create)), a browser with Web Audio + WebGL (any modern desktop browser), and a microphone if you want to clone a voice or try the mic lab. No code is required for the labs; the live-agent step assumes light JavaScript familiarity.
+[![The Voice Lab page with the voice catalogue and preview controls](/docs/img/voice-lab.webp)](/voice)
+
+*Voice Lab previews every voice the speech and lipsync stack can drive.*
+
---
## What you're building
diff --git a/docs/tutorials/walk-companion.md b/docs/tutorials/walk-companion.md
index 6b01ad3ff..306e743da 100644
--- a/docs/tutorials/walk-companion.md
+++ b/docs/tutorials/walk-companion.md
@@ -6,6 +6,14 @@ Along the way you'll see what the [/walk](/walk) product offers, live demos acro
**Prerequisites:** a website you can edit and a WebGL-capable browser. The iframe path needs no tooling at all; the npm path assumes light JavaScript familiarity and that you already bundle [Three.js](https://threejs.org) (>= 0.150).
+[![The Walk experience on a phone-sized viewport with the avatar on screen](/docs/img/walk-mobile.webp)](/walk)
+
+*Walk on a phone: the same rig, the same clips, thumb-sized controls.*
+
+[![Animated loop of a three.ws avatar walking in the Walk experience](/docs/img/walk-loop.webp)](/walk)
+
+*Captured live: the default rig running the canonical walk clip.*
+
---
## What you're building
diff --git a/public/docs/img/scene-studio.webp b/public/docs/img/scene-studio.webp
new file mode 100644
index 000000000..d8de35dd5
Binary files /dev/null and b/public/docs/img/scene-studio.webp differ
diff --git a/public/docs/media-manifest.json b/public/docs/media-manifest.json
index 35ac91864..5c784bf12 100644
--- a/public/docs/media-manifest.json
+++ b/public/docs/media-manifest.json
@@ -1,6 +1,6 @@
{
"generatedBy": "scripts/capture-doc-media.mjs",
- "generatedAt": "2026-07-31T19:25:22.123Z",
+ "generatedAt": "2026-09-08T17:25:34.474Z",
"shots": {
"animation-studio": {
"id": "animation-studio",
@@ -180,6 +180,23 @@
"width": 1800,
"height": 1125
},
+ "scene-studio": {
+ "id": "scene-studio",
+ "src": "/docs/img/scene-studio.webp",
+ "alt": "Scene Studio with a 3D scene in the viewport and the object rail",
+ "caption": "Scene Studio composes objects and avatars into one shareable scene.",
+ "route": "/scene",
+ "viewport": "desktop",
+ "animated": false,
+ "authenticated": false,
+ "capturedAt": "2026-09-08T17:25:32.557Z",
+ "capturedFrom": "https://three.ws",
+ "commit": "7f812445c",
+ "sha256": "e81750217c6d3850efe7d9a0e73f135599d8a83935d7b33e27f6193a1a8ad96c",
+ "bytes": 59536,
+ "width": 1800,
+ "height": 1125
+ },
"sign-api-console": {
"id": "sign-api-console",
"src": "/docs/img/sign-api-console.webp",