design-builder

Design-to-code pipeline: discover, extract, tokenize, build.

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "design-builder" with this command: npx skills add adeonir/agent-skills/adeonir-agent-skills-design-builder

Design Builder

Design-to-code pipeline: discover, extract, tokenize, build.

Workflow

discovery --> copy --> design --> frontend / variants / export

Each step is independent. Can run isolated or chained. Discovery is always the first step -- never skip it.

Context Loading Strategy

Load only the reference matching the current trigger. For frontend and variants operations, also load aesthetics.md and web-standards.md as auto-loaded dependencies.

Never simultaneous:

  • Multiple operation references (e.g., copy.md + frontend.md)

Triggers

Extraction

Trigger Pattern Reference

Extract copy, copy from URL, content from website copy.md

Extract design, design from image, design tokens design.md

Building

Trigger Pattern Reference

Build frontend, create components, generate React frontend.md

Generate variants, preview designs, HTML variants variants.md

Export design, export to Figma, send to Figma export.md

Auto-Loaded (not direct triggers)

  • aesthetics.md -- loaded by frontend.md and variants.md as design principles

  • web-standards.md -- loaded by frontend.md and variants.md as implementation rules

Cross-References

copy.md ---------> design.md (content informs design) design.md -------> frontend.md (tokens required) design.md -------> variants.md (tokens required) aesthetics.md ------> frontend.md (design principles) aesthetics.md ------> variants.md (design principles) web-standards.md --> frontend.md (implementation rules) web-standards.md --> variants.md (implementation rules) variants.md -----> frontend.md (user picks variant, then builds React) variants.md -----> export.md (variants required for Figma export)

Discovery

Before any operation, establish project context.

Step 1: Check Existing Context

Look for existing documents in .artifacts/docs/ :

  • prd.md -- PRD

  • brief.md -- Brief

If found: read and extract purpose, audience, tone, and key features. Skip to the relevant trigger operation.

Step 2: Lightweight Discovery (when no PRD/Brief exists)

Ask up to 4 questions, one stage only:

  • What is the project purpose? (landing page, app, tool, portfolio)

  • Who is the target audience?

  • What is the visual reference? (URLs, screenshots, descriptions)

  • Any brand or style constraints? (colors, fonts, existing guidelines)

If the user answers "I don't know" to any question, mark as TBD and move forward. Summarize understanding before proceeding.

Step 3: Route to Operation

Phase 1 -- Extraction (how to obtain design tokens):

Has URL reference? Yes --> Extract copy --> Extract design No --> Has image reference? Yes --> Extract design No --> Visual discovery (tone, colors, typography) --> Extract design

Phase 2 -- Building (what to build -- user chooses):

design.json exists --> What to build? Preview first --> Variants --> Frontend or Export Build directly --> Frontend Send to Figma --> Variants --> Export External tool --> Generate prompt (v0, aura.build, replit, etc.)

Valid paths after design.json:

  • design --> variants --> frontend

  • design --> variants --> export

  • design --> frontend (directly)

  • design --> prompt for external tool

Templates

Context Template

Copy extraction output copy.md

Design tokens output design.md

Artifacts

.artifacts/design/ ├── copy.yaml # Structured content ├── design.json # Design tokens └── variants/ ├── minimal/index.html # Variant preview ├── editorial/index.html ├── startup/index.html ├── bold/index.html ├── {custom}/index.html # Custom variant (if requested) └── index.html # Comparison page src/ # React components (frontend)

Guidelines

DO:

  • Ask user for project name and use kebab-case for directory names

  • Check for existing PRD/Brief before any operation and use them as context

  • Check for existing copy.yaml, design.json before starting

  • Suggest next steps after completing any operation (defined in each reference file)

  • Suggest missing prerequisites (e.g., "Run extract design first to generate design.json")

DON'T:

  • Skip discovery -- always establish project context first

  • Ignore existing artifacts when they're available

  • Couple suggestions to specific skills (keep them generic)

  • Block on missing PRD/Brief -- run lightweight discovery instead

Output

.artifacts/design/copy.yaml # Structured content from URL extraction .artifacts/design/design.json # Design tokens from image extraction .artifacts/design/variants/ # HTML+CSS preview variants src/ # React components from frontend building

Error Handling

  • No PRD/Brief: Run lightweight discovery, never block on it

  • No copy.yaml: Proceed without it, or suggest running extract copy first

  • No design.json: Required for frontend/variants/export -- suggest running extract design

  • Reference URL unavailable: ask user to paste a screenshot instead

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

Automation

spec-driven

No summary provided by upstream source.

Repository SourceNeeds Review
Automation

git-helpers

No summary provided by upstream source.

Repository SourceNeeds Review
Automation

debug-tools

No summary provided by upstream source.

Repository SourceNeeds Review
Automation

integrating-jupiter

No summary provided by upstream source.

Repository SourceNeeds Review