3d Assembly Manual logo

3d Assembly Manual

Community
SkylarKitchen
3d-assembly-manual

Build a phone-first interactive 3D step-by-step guide (LEGO-manual style, BIG diagram look) as a single self-contained Three.js HTML file. Use when the user asks for a 3D build guide, install sequence, assembly/instruction manual, or "step-by-step 3D" for a physical task (garden, house, furniture, repair).

Overview

PublisherSkylarKitchen
Repositoryskills
Skill name3d-assembly-manual
Stars
132
Forks
16
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 SkylarKitchen on GitHub. Read the source before you install it.

Installation

Install the 3d Assembly Manual 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/SkylarKitchen/skills/tree/main/skills/design/3d-assembly-manual
  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/SkylarKitchen/skills.git /tmp/skills
mkdir -p .claude/skills
cp -r /tmp/skills/skills/design/3d-assembly-manual .claude/skills/3d-assembly-manual
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable 3d Assembly Manual 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 3d Assembly Manual 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 3d Assembly Manual 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.

3D assembly manual

Proven on a garden-bed build, then generalised after kitchen and sauna test builds. Copy starter.html: a project-agnostic engine with a four-step shelf example that renders on its own. Edit only the CONFIG block and the PROJECT SCENE block; leave the engine alone.

CONFIG keys (top of the script)

KeyWhat
title, subtitle, dateHeader text; date prefixes the page counter.
storageKeylocalStorage key for the last page; unique per manual.
accent / fillDark accent (arrows, plan frame, outline flash, CSS --accent) and pale part fill (this step's parts, thumbnails).
viewDefault view direction vector.
planPlan inset {c:[x,z], r}; null derives it from the scene's Box3 (shadow camera always does).
figureOff (null) by default; opt in only when asked, with [x, z, rotY?]: a flat cut-out that faces the view unless rotY is given. Frame the overview preset to take in its raised hand, 7.4 ft up.
CAMPresets {c, r, v?, walls?, cut?}.
NAMES, INVENTORYPart labels keyed like P; inventory groups [[group, [[part, qty]]]].
STEPSPages. Besides the copy keys (title text acts meta time tip parts dims notes inventory calendar): detail {p, r, hide?:[regKeys], view?} (the bubble's own pass: hide parts in front, look from another direction), view (direction), walls {h, sides?} (walls on those sides drop to h ft; omit sides = all), cut:[regKeys] (fade to 0.12 for a cutaway), onEnter(t) (special animation; t has lastLand, objs, tween, ease, easeIO, setOpacity, setColor, paint, T, still; return a new end time to delay marks). Step keys override the preset.

PROJECT SCENE block: context geometry (auto-painted white), wall(side, x0, x1, y0, y1, z0, z1) for rooms, P.* part builders, optional THUMB_ONLY[key] builders (a thicker thumbnail-only variant for thin parts: bands, wires), reg() calls. Groups passed to reg() are re-centred automatically, so parts can be built in absolute coordinates.

Look (BIG evolution diagram)

  • White model, ink hairline edges (EdgesGeometry, opacity ~0.72), near-flat light: HemisphereLight 0.9 + sun 0.16, one pale PCFSoft shadow. Toon gradient bands 226/241/255.
  • Already-built parts near-white (#ececec). Only this step's new or removed parts in the pale fill; arrows in the dark accent. Parts page shows the whole kit in fill. The starter default is the blue pair #1F4FD8 / #A6C0F6; one orange (#f28c1e) for both also works.
  • No people in the scene by default: every figure tried so far (cylinder people, then a flat Modulor cut-out) read as weird. Scale comes from the dimension lines and the live scale bar. modulorFigure() stays in the starter only for an explicit request. Chrome neutral: white paper, ink buttons, grey rules, Geist + Geist Mono, sentence case, square corners. Colour lives only in the picture.

Engine (in starter.html)

  • reg(key, obj, {add, remove, only:[pages], from, lag, each, span, part, grow, noGhost, noArrow}); a page's scene is the cumulative state; only parts fade when their page turns.
  • Orthographic camera; preset {c, r, v} tweened 900 ms ease-in-out, walls and cutaways tweened with it. Every tween's clock starts on its first rendered frame, so a slow first frame can't eat the opening motion. Landings 700 ms ease-out along an arrow, ghost outline at destination, accelerating stagger span*sqrt(i/(n-1)).
  • Per step: parts callout with 3D thumbnails (second renderer, toDataURL; below 760 px one scrolling row of 32 px thumbs, ellipsised names, right-edge fade when it overflows), numbered action marks with screen offsets (clamped below the callout; the leader is redrawn from the clamped position to its point), detail bubble (scissor pass, circular mask with overflow:hidden), dimension lines, live scale bar, plan inset, step number, time, one tip.
  • Tap a part = blink + outline flash. Swipe the sheet (pointer events, touch-action:pan-y). #stepN deep-links, #stepN&still freezes for captures.

Process

  1. Measure or estimate the real space from photos (mark "verify"); list parts grouped by where they come from.
  2. Write STEPS copy: plain instructions, 2-4 numbered acts, quantities in a mono meta line.
  3. Build (the starter has no dark-mode tokens: it commits to a light drawing sheet on purpose), then verify at 375 px in a browser over a localhost server (e.g. python3 -m http.server; keep the viewport meta), and capture with headless Chrome through a 390x844 iframe harness at #stepN&still.
  4. Deliver one .html file: Three.js and fonts load from a CDN, everything else is inline, so it opens by double-click or on any static host. Until you have a capture from the browser the reader will use, call the render unverified.

Gotchas

  • MeshToonMaterial in shadow drops the whole directional term: keep ambient high or orange goes brown.
  • Keep pulse base colour under its own userData key (opacity code owns userData.base).
  • Emissive glow is invisible on white or accent parts; blink opacity instead.
  • Touch-only handlers can't be tested with mouse-driven automation; use pointer events.
  • Rooms: work on opposite walls (sink run vs nook) needs two presets with different v, each cutting the walls between camera and work (learned on a kitchen build). The pale fill matters: a dark part fill hides the ink edges.
  • Numbers derived from another file (plank rows, rips) : port its calculation functions instead of retyping results.
  • Captures: run headless Chrome one at a time, each under perl -e 'alarm 50; exec @ARGV' <chrome --headless=new --user-data-dir=<scratch>/st-N --disable-gpu --use-angle=swiftshader --enable-unsafe-swiftshader --window-size=520,880 --timeout=15000 --screenshot=...> (macOS has no timeout; it does not exit on its own), then pkill -f user-data-dir=<scratch>/st-N. Parallel runs give black canvases.

Sources

  • Video with no transcript (music-only): read the burned-in captions from frames. Draw the paused video to a canvas at intervals, lay the frames out as a contact sheet on the page, screenshot it. Seeking in a background tab renders black.

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 3d Assembly Manual AI skill do?

Build a phone-first interactive 3D step-by-step guide (LEGO-manual style, BIG diagram look) as a single self-contained Three.js HTML file. Use when the user asks for a 3D build guide, install sequence, assembly/instruction manual, or "step-by-step 3D" for a physical task (garden, house, furniture, repair).

Why use 3d Assembly Manual on TypingMind?

Because you install it once and use it with any model. 3d Assembly Manual 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 3d Assembly Manual in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/SkylarKitchen/skills/tree/main/skills/design/3d-assembly-manual. 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 3d Assembly Manual?

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 3d Assembly Manual?

As many as you like. As long as a model supports skills, you can use 3d Assembly Manual with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the 3d Assembly Manual AI skill free?

Yes. It is published on GitHub by SkylarKitchen 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 👇