Design Engineering logo

Design Engineering

Organization
AgentsORG
design-engineering

Load when reviewing UI code, designing a component or layout, picking an easing curve or transition, deciding whether something should animate or make a sound, choosing typography, color (OKLCH, contrast) or avatars, writing UI copy or errors, auditing for AI-default tells, code tells or a11y misses, making a screen look less generated, consuming DESIGN.md or .design tokens, writing design-system docs for agents, giving Agentation feedback, asking why a UI feels flat or unfinished, judging when delight earns its weight, designing or generating UI sound effects and syncing them to motion, planning, cutting, scoring or reviewing a product launch video, creating or animating SVG (icons, mascots, logo reveals, video-to-vector), drawing a diagram or tldraw canvas, prototyping directions behind a picker, building a tool instead of re-prompting, or choosing which design skill should own a job. Distills Emil Kowalski, Benji Taylor, transitions.dev, guidelines.sh, Vercel, Index, Apple and 47 measured launch films.

Overview

PublisherAgentsORG
Repositorydesign-engineering
Skill namedesign-engineering
Stars
35
Forks
2
Bundled files
167
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.

  • 167 bundled files

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

  • Open source

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

Installation

Install the Design Engineering 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/AgentsORG/design-engineering/tree/main/skills/design-engineering
  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/AgentsORG/design-engineering.git /tmp/design-engineering
mkdir -p .claude/skills
cp -r /tmp/design-engineering/skills/design-engineering .claude/skills/design-engineering
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Design Engineering 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 Design Engineering 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 Design Engineering 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.

design-engineering

This skill is a graph, not a file. The body below is the Map of Content. Follow the [[wikilinks]] only as needed — that's the whole point of progressive disclosure.

How to use this skill

  1. /design-engineering is a router before it is a library: [[skill-router]] resolves the design contract, classifies the phase (undecided → build → refine → check → name), and hands the job to one owner — a node here, a subagent, or an installed companion skill (AgentsORG design, impeccable, HyperFrames, ElevenLabs, transitions-dev, the shadcn CLI, diagram-design, tldraw, GSAP, better-icons, Jakub Krehel's interfaces skills, fun-isometric). One or two owners, never five.
  2. Route first: [[routing-table]] maps intent → entry node for most single questions. If two intents blur, [[disambiguation]] names the tiebreaker. Multi-cluster jobs follow [[stacking-chains]].
  3. Otherwise scan the MOCs below, pick the cluster, and open its MOC-*.md for the atomic nodes under it.
  4. Read only the atomic nodes that match. Each node is standalone — you don't need siblings.
  5. Before reviewing UI code, also load [[gotchas]] and [[pov]].

If the user asks for a UI code review, use the format defined in [[review-format]] and scan against [[review-checklist]].

Philosophy — when, where, and why polish matters

How to think about taste, delight, and the difference between marketing and product UI.

  • [[MOC-philosophy]] → [[taste-is-trained]], [[unseen-details-compound]], [[beauty-is-leverage]], [[delight-impact-curve]], [[feeling-right]], [[marketing-vs-product-ui]], [[states-are-the-work]], [[data-is-content]], [[dependency-discipline]], [[pointing-beats-describing]], [[articulate-precisely]]

Motion — when something should move, and how

The largest cluster. Animation is the most overused tool in modern UI; this cluster tells you when not to animate as much as when to.

  • [[MOC-motion]] → [[animation-decision-framework]], [[easing-curves]], [[duration-table]], [[spring-animations]], [[transform-opacity-only]], [[performance-discipline]], [[transform-mastery]], [[clip-path-tricks]], [[never-scale-from-zero]], [[gesture-momentum]], [[stagger-choreography]], [[prefers-reduced-motion]], [[fly-not-teleport]], [[responsive-feedback]], [[sonner-principles]], [[debugging-animations]], [[lerp-breathing]], [[morphing-icons]], [[shared-letter-morph]], [[hover-default-imperative]], [[tray-rules]], [[css-conventions]], [[frame-rate-independence]], [[animated-hit-testing]]

Transition techniques — Jakub Antalik's catalog

A subset of motion focused on canonical transitions for common UI archetypes. Where the Motion MOC teaches principles, these nodes teach implementation. From transitions.dev.

  • [[cross-blur-transitions]] — Pair opacity 0↔1 with filter: blur(2px) ↔ 0 to mask imperfect crossfades.
  • [[compose-subtract-asymmetry]] — Enter with more properties than exit. Disappearance feels soft, not reversed.
  • [[distance-falloff-propagation]] — Per-element lift via lift * pow(falloff, distance) for grouped hover.
  • [[multi-segment-shake]] — Form-error shake at 0%, 28.57%, 57.14%, 78.57%, 100% over 280ms.

Sound — when an interface should be heard, and how

The sense the web forgot. Product UI is silent by default; a launch video is the reverse. This cluster decides which, designs one material family, syncs it to motion, and generates the files — ElevenLabs on demand, or open-weight / procedural / CC0 without a key.

  • [[MOC-sound]] → [[sound-decision-framework]], [[sound-motion-sync]], [[sound-palette]], [[sound-spec]], [[sound-playback-web]], [[sound-generation-elevenlabs]], [[sound-generation-open-source]], [[launch-video-sound]], [[launch-video-music]], [[sound-from-motion]], [[designed-sfx-anatomy]], [[earshot-reconciliation]], [[sound-licensing]]

Launch video — planning, cutting, scoring and reviewing a launch film

A rendered film is its own surface: name the register, time the beats, set the words and the UI demo, make the seams and cuts, place the sound, review against the register. HyperFrames builds; this cluster decides the register, beats, words and sound, and [[hyperframes-reconciliation]] settles seams inside a HyperFrames project.

  • [[MOC-launch-video]] → [[launch-video-registers]], [[launch-video-structure]], [[launch-video-type]], [[launch-video-ui-demo]], [[launch-video-motion]], [[launch-video-seams]], [[launch-video-cuts]], [[launch-video-review]], [[hyperframes-reconciliation]], [[launch-video-formats]], [[cinetic-reconciliation]]

SVG — creating, animating, morphing, and vectorizing

The only image format that is also an interface. Author it on the token system, move it with the engine its home allows, morph it by the command-count rule, and turn flat clips into editable mascots.

  • [[MOC-svg]] → [[svg-creation]], [[svg-animation]], [[svg-path-morphing]], [[video-to-vector-pipeline]], [[isometric-stickers]], [[fun-isometric-reconciliation]]

Diagrams — figures, canvases, and whether to draw at all

A diagram is the costliest way to say something. Decide whether a picture earns its place and who touches it next. Then draw a finished figure (Cathryn Lavery's diagram-design) or edit a live canvas (tldraw's Desktop skill or MCP app). Map the project's contract onto the figure, and review it by the connector rules.

  • [[MOC-diagram]] → [[diagram-decision]], [[diagram-craft]], [[diagram-design-reconciliation]], [[tldraw-canvas]]

Typography — humanizing text

The font defaults of the AI era are tells. Better choices and the rules around them.

  • [[MOC-typography]] → [[typography-humanity]], [[line-length-tracking]], [[type-scale-and-rhythm]], [[line-behavior]]

Surface — color, shadow, radius, dark mode, imperfection

The "background" choices that most decks of guidelines skip. They're load-bearing for feel.

  • [[MOC-surface]] → [[color-monochromatic]], [[color-scales-oklch]], [[dark-mode]], [[shadows-whisper]], [[depth-and-nesting]], [[border-radius]], [[visual-imperfection]], [[contrast-and-color-scheme]], [[z-index-strategy]], [[color-token-tiers]], [[palette-step-roles]], [[gradient-interpolation]]

Layout — page-level grids, viewports, sticky chrome, URL-as-state, marketing surfaces

The container, not the component. Marketing especially benefits from custom-per-viewport thinking; product apps benefit from URL-driven state.

  • [[MOC-layout]] → [[viewport-custom-design]], [[sticky-and-scroll-tells]], [[url-as-state]], [[marketing-surface-rules]], [[page-frame]], [[scroll-anchoring]], [[grouping-by-space]], [[growth-and-overflow]], [[logical-properties-rtl]]

Components — buttons, hovers, states, cards, forms, avatars, icons, a11y, copy, APIs

The atoms users actually touch. Their behavior is where craft lives.

  • [[MOC-components]] → [[hover-states-subtle]], [[empty-loading-states]], [[icon-systems]], [[cards-design]], [[forms-validation]], [[forms-behavior]], [[touch-and-focus]], [[ui-polish-pass]], [[component-api-design]], [[avatar-systems]], [[interaction-personality]], [[accessibility-baseline]], [[optimistic-updates]], [[copy-voice]], [[dashboard-tells]], [[screen-reader-announcements]], [[keyboard-patterns]], [[wcag-thresholds]], [[strings-and-locale]], [[destructive-actions]]

Anti-patterns — what AI-generated UI gives away

What looks "AI default" and what to do instead. High-value because it's about deletion, not addition.

  • [[MOC-anti-patterns]] → [[ai-default-tells]], [[unslop-pass]], [[content-authenticity]], [[copy-tells]], [[code-tells]], [[screenshot-critique]]

Meta — review format, design.md, Agentation, evals, per-installer files

Procedural rules and growing files. [[review-format]] is mandatory when doing UI code reviews.

  • [[skill-router]] — What /design-engineering does first: resolve the contract, classify the phase, hand off to one owner — a node, a subagent, or an installed companion (AgentsORG design, impeccable, HyperFrames, ElevenLabs, transitions-dev, shadcn CLI, diagram-design, tldraw, GSAP, better-icons, fun-isometric).
  • [[routing-table]] — Intent → entry-node router with the four postures (build / judge / decide / name). Open this before anything else on a single question.
  • [[prototype-picker]] — Undecided? Three to five genuinely different versions behind a live picker, numbers on controls, then write the decision and delete the harness.
  • [[build-a-tool]] — "Closer, but not quite" twice? Stop re-prompting; diagnose the gap and build the small tool that produces the artifact.
  • [[vibe-to-generator]] — Only a mood, no spec? Research the visual language into rules, then a seeded generator with knobs and a save button.
  • [[design-system-docs]] — Writing a DESIGN.md or docs folder an agent can follow: inventory the code, when/alternative, closed variant sets, correct/incorrect examples, verify by prompting.
  • [[skill-writing-rules]] — Encode process, write the why, be strict, earn every line, one aspect per skill, test by running. Governs every node here and [[pov]].
  • [[design-benchmarks]] — DesignBench's generation / edit / repair tasks and Design Arena's pairwise votes, and this skill's own eval suite that borrows both.
  • [[disambiguation]] — The questions that blur together (four motion questions, three dark-mode owners, two layout-shift owners) and which node owns each.
  • [[stacking-chains]] — Ordered node chains for multi-step jobs (new screen, feel-better pass, design system, marketing site, a11y pass).
  • [[review-format]] — Required output format for UI critiques (Before | After | Why markdown table).
  • [[review-checklist]] — Twenty-six-row audit to scan against before signing off on a UI review.
  • [[change-scoped-review]] — Reviewing a diff, not a screen: scope, blast radius, the removed side of every hunk, and each finding statused Introduced, Regression or Pre-existing.
  • [[stress-scenarios]] — One component under every scenario that can reach it (length, shape, quantity, container, state) on a throwaway harness, and its product states behind a switcher.
  • [[reading-others-interfaces]] — How someone else's site or effect was built: evidence not instruction, measured / derived / inferred, layers in paint order, the recipe in words.
  • [[interfaces-reconciliation]] — Jakub Krehel's interfaces skills as companions: who owns which job when installed, and twelve value conflicts settled.
  • [[design-vocabulary]] — The precise word for each design phenomenon and where the skill goes deeper. The lexicon behind [[articulate-precisely]]; seeded by Index (Emil Kowalski & Glenn Carstens-Peters).
  • [[using-design-file]] — How to consume a project's .design contract (schema design.v1). Machine-readable tokens, committed intent, executable policy — and it outranks this skill's defaults.
  • [[using-design-md]] — How to consume a project's DESIGN.md (Google Labs spec). Tokens are normative; prose is context.
  • [[agentation-workflow]] — Install and use Agentation for click-to-annotate design review. Two-session critique-then-fix workflow.
  • [[cross-model-testing]] — How to test description / routing changes across GPT, Claude Opus, and Claude Sonnet.
  • [[gotchas]] — lived failures, appended as the agent trips up.
  • [[pov]] — author/installer's opinions and taste calls that override defaults. Edit this when you fork.
  • [[animations-dev-curriculum]] — external pointer to Emil's course; don't duplicate.

Workflow subagents — sibling agents/ directory

Ten narrow-purpose subagents live in agents/ next to this file. Spawn one when its specific workflow matches the user's ask. Each subagent loads its own slice of the graph and returns to the main agent with a structured result.

  • [[ui-reviewer]] — Runs the [[review-format]] table + [[review-checklist]] audit on a UI snippet.
  • [[motion-auditor]] — Animation/transition critique against the motion cluster + transitions.dev nodes.
  • [[anti-pattern-scanner]] — Scans for AI-default tells and content-authenticity misses.
  • [[agentation-fix-loop]] — Session-2 fix side of [[agentation-workflow]]. Reads MCP annotations and applies fixes.
  • [[design-md-consumer]] — Reads a project's DESIGN.md and threads its tokens through generated UI per [[using-design-md]].
  • [[pov-curator]] — Helps the installer fork [[pov]] and append to [[gotchas]] after a real failure.
  • [[sound-designer]] — Runs the sound cluster end to end: decide, palette, generate (scripts/sound-family.mjs for a product family, scripts/sound-sheet.mjs for a video stem derived from the motion), spec, wire up. Returns a sound map.
  • [[svg-creator]] — Authors or refactors an SVG asset: grid, named groups, token colors, SVGO with the right flags, accessible name.
  • [[svg-animator]] — Animates an SVG or builds a vector flipbook from frames (scripts/svg-flipbook.mjs), engine chosen by where the file lives.
  • [[launch-film-analyst]] — Measures a reference launch film or your own render at native fps and returns the [[launch-video-review]] rubric plus a per-film JSON.

See agents/README.md for the full directory and selection guide.

Identity — SOUL.md

The repo also ships a SOUL.md at root — voice, stance, and taste lineage for any agent operating in this skill. AGENTS.md is the "what/how"; SOUL.md is the "who/why". Read it once per session if your harness doesn't auto-inject it.

Tax check

Every sentence above costs tokens in every session. If you find yourself adding a sentence, ask: would the agent get this wrong without it? If no, delete it. If yes, it belongs in an atomic node, not here.

— HKTITAN, 2026.

Bundled files

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

and 107 more files.

Frequently asked questions

What does the Design Engineering AI skill do?

Load when reviewing UI code, designing a component or layout, picking an easing curve or transition, deciding whether something should animate or make a sound, choosing typography, color (OKLCH, contrast) or avatars, writing UI copy or errors, auditing for AI-default tells, code tells or a11y misses, making a screen look less generated, consuming DESIGN.md or .design tokens, writing design-system docs for agents, giving Agentation feedback, asking why a UI feels flat or unfinished, judging when delight earns its weight, designing or generating UI sound effects and syncing them to motion, pl...

Why use Design Engineering on TypingMind?

Because you install it once and use it with any model. Design Engineering 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 Design Engineering in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/AgentsORG/design-engineering/tree/main/skills/design-engineering. 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 Design Engineering?

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 Design Engineering?

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

Is the Design Engineering AI skill free?

Yes. It is published on GitHub by AgentsORG 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 👇