Token pipeline (Style Dictionary v4, DTCG format): - Primitive tokens: colour palettes (brand, sage, neutral, feedback), typography (3 font families, 21-variant type scale), spacing (4px grid), border radius, shadows, opacity - Semantic tokens: text, surface, border, interactive, feedback colours; typography roles; layout spacing - Component tokens: Button (4 sizes), Input (2 sizes) - Generated outputs: CSS custom properties, JS ES6 module, flat JSON Atoms (3 components): - Button: contained/soft/outlined/text × primary/secondary, 4 sizes, loading state, underline for text variant - Typography: 21 variants across display/heading/body/label/caption/overline, maxLines truncation - Input: external label, helper text, error/success validation, start/end icons, required indicator, 2 sizes, multiline support Infrastructure: - MUI v5 theme with full token mapping - Storybook 8 with autodocs - Claude Code agents and skills for token/component workflows - Design system documentation and cross-session memory Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
3.2 KiB
3.2 KiB
FA Design System — Quick Start
What you've got
A complete Claude Code project scaffold with:
- 3 agents: token-architect, component-builder, story-writer
- 6 slash commands: /create-tokens, /build-atom, /build-molecule, /sync-tokens, /status, /review-component
- 4 memory files: session-log, decisions-log, component-registry, token-registry
- 2 convention docs: token-conventions, component-conventions
- Living design system spec: docs/design-system.md
- MCP config: Figma remote MCP pre-configured
- Full React + MUI + Storybook + Style Dictionary setup
Setup (5-10 minutes)
# 1. Extract the project
tar -xzf fa-design-system-scaffold.tar.gz
cd fa-project
# 2. Run the bootstrap script
chmod +x bootstrap.sh
./bootstrap.sh
# 3. Set up Figma MCP (in Claude Code)
claude mcp add --transport http figma-remote-mcp https://mcp.figma.com/mcp
# Restart Claude Code, then /mcp → authenticate with Figma
# 4. Start Storybook (separate terminal)
npm run storybook
Your workflow
Step 1: Create tokens
claude
/create-tokens I want to create the FA design system. Here are my brand colours: [paste hex values or attach reference images]. Fonts: [your font choices]. The platform serves Australian families planning funerals — warm, trustworthy, calm aesthetic.
Step 2: Build atoms (one at a time)
/build-atom Button
# Review in Storybook at http://localhost:6006
# Provide feedback, iterate
/build-atom Input
/build-atom Typography
/build-atom Card
# ... etc
Step 3: Build molecules
/build-molecule PriceCard
/build-molecule FormField
# ... etc
Check status anytime
/status
Review a component
/review-component Button
How memory works
Every agent reads these files before starting work and updates them after:
| File | Purpose |
|---|---|
docs/memory/session-log.md |
What happened last session, what's next |
docs/memory/decisions-log.md |
Every design decision with rationale |
docs/memory/component-registry.md |
Status of every component |
docs/memory/token-registry.md |
All tokens with values and usage |
This means you can close Claude Code, come back tomorrow, and the agents will pick up exactly where you left off.
File structure
fa-project/
├── CLAUDE.md ← Claude Code reads this every session
├── .claude/
│ ├── agents/ ← Agent definitions
│ ├── commands/ ← Slash commands
│ └── skills/ ← Shared knowledge
├── .mcp.json ← MCP server config
├── docs/
│ ├── memory/ ← Cross-session memory
│ ├── conventions/ ← Rules agents follow
│ ├── reference/ ← Setup guides
│ └── design-system.md ← Living design spec
├── tokens/ ← DTCG token JSON (source of truth)
├── style-dictionary/ ← Token build config
├── src/
│ ├── theme/ ← MUI theme + generated CSS vars
│ └── components/ ← Atoms, molecules, organisms
└── .storybook/ ← Storybook config with theme provider