本文へ移動
cccskills
無料GitHub で公開

design-md

Create, fetch, validate, or export DESIGN.md design system files in the Google Stitch 9-section format. Fetch reference designs from 68 brands (Linear, Stripe, Vercel, etc.) via getdesign CLI, generate scaffolds from existing CSS custom properties and Tailwind tokens, or validate and export with the Google @google/design.md CLI. Pairs with minoan-frontend-design for implementation. Triggers on "DESIGN.md", "create design system file", "extract design tokens", "codify the design", "getdesign", "make it look like [brand]", "lint design", "validate DESIGN.md", "export design tokens", "export tailwind theme".

インストール方法を見る

含まれるファイル(7)

  • SKILL.md6.8 KB
  • .gitignore74 B
  • README.md324 B
  • references/design-md-format.md9.8 KB
  • references/google-cli.md3.2 KB
  • scripts/fetch_all_design_md.sh1.8 KB
  • scripts/generate_design_md.py18.7 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

DESIGN.md — Design System Library & Generator

DESIGN.md is a plain-text design system document in the Google Stitch 9-section format that AI agents read to generate consistent UI. It is the token layer between design intent (.design-context.md from /shape) and implementation (/minoan-frontend-design).

Three Workflows

Workflow A: Fetch a reference DESIGN.md

Use the getdesign CLI to fetch from a library of 68 brand design systems:

npx getdesign@latest add <brand>        # Drops DESIGN.md into project root
npx getdesign@latest add stripe --out ./docs/DESIGN.md  # Custom output path
npx getdesign@latest list               # Show all 68 available brands

Available brands include: airbnb, apple, claude, cursor, figma, framer, linear.app, nike, notion, shopify, spotify, stripe, supabase, tesla, vercel, and 53 more. Each is a ~350-line file with exact hex values, font stacks, component specs, shadow systems, and responsive breakpoints.

After fetching, /minoan-frontend-design reads the DESIGN.md during implementation (its QA normalize step aligns output with project DESIGN.md tokens).

Pre-fetched Library

All 68 brand DESIGN.md files can be pre-fetched into a local reference library:

bash ~/.claude/skills/design-md/scripts/fetch_all_design_md.sh         # Fetch all 68
bash ~/.claude/skills/design-md/scripts/fetch_all_design_md.sh --force # Re-fetch

The library lives at ~/.claude/skills/design-md/library/<brand>/DESIGN.md. Read any brand directly without running npx:

cat ~/.claude/skills/design-md/library/stripe/DESIGN.md

The library/ directory is gitignored — it's a local cache, not distributed with the skill.

Workflow B: Generate a custom DESIGN.md

Extract design tokens from an existing project's CSS and scaffold a DESIGN.md:

python3 ~/.claude/skills/design-md/scripts/generate_design_md.py <project-path>
python3 ~/.claude/skills/design-md/scripts/generate_design_md.py . --output DESIGN.md

The script scans for CSS files containing @theme blocks (Tailwind CSS 4), :root custom properties, font declarations, shadows, border radii, and breakpoints. It organizes extracted tokens into the 9-section format with TODO placeholders for prose sections (Visual Theme, Do's/Don'ts) that require human or Claude completion.

After generating the scaffold, fill in the prose sections using project context — the conceptual direction, the atmosphere, the design guardrails. The format spec is at references/design-md-format.md.

Workflow C: Validate and export with Google CLI

Use @google/design.md (the upstream CLI for the 9-section format) to lint, diff, and export:

npx @google/design.md lint DESIGN.md                       # 8 validation rules, JSON output
npx @google/design.md diff OLD.md NEW.md                   # Token-level diff, exits 1 on regressions
npx @google/design.md export --format tailwind DESIGN.md   # Tailwind theme JSON
npx @google/design.md export --format dtcg DESIGN.md       # W3C Design Token Format JSON

Lint after generate: Run lint after Workflow B to surface missing sections, broken refs, and contrast issues before prose completion.

Lint before handoff: Run lint before passing DESIGN.md to /minoan-frontend-design. Clean lint (exit 0) confirms structural completeness.

Export for implementation: export --format tailwind produces a theme JSON that downstream tooling can consume directly.

Diff for versioning: Compare a project's DESIGN.md against a library reference or audit changes between revisions. Exit 1 = regressions (removed tokens, broken contrast).

See references/google-cli.md for lint rule reference and exit code semantics.

The 9-Section Format

#SectionContent
1Visual Theme & AtmosphereProse: mood, philosophy, conceptual direction
2Color Palette & RolesGrouped: semantic name + hex + functional role
3Typography RulesFont families, hierarchy table (size/weight/tracking/line-height)
4Component StylingsButtons, cards, inputs, nav — CSS values + states
5Layout PrinciplesSpacing scale, grid, whitespace philosophy, border-radius scale
6Depth & ElevationShadow system table, surface hierarchy
7Do's and Don'tsDesign guardrails and anti-patterns
8Responsive BehaviorBreakpoints table, touch targets, collapsing strategy
9Agent Prompt GuideQuick color reference for copy-paste

See references/design-md-format.md for the full annotated spec with examples for each section.

Pipeline Integration

/shape → .design-context.md (intent: audience, mood, anti-goals)
/design-md → DESIGN.md (tokens: hex values, font stacks, shadow systems)
  lint → npx @google/design.md lint DESIGN.md (validation gate)
/minoan-frontend-design → implementation (creative direction + materials)
/design-audit + /design-polish → refinement

.design-context.md and DESIGN.md are complementary:

  • .design-context.md captures who and why — audience, aesthetic direction, design dials (DESIGN_VARIANCE, MOTION_INTENSITY, VISUAL_DENSITY), anti-goals
  • DESIGN.md captures what exactly — hex colors, font-size/weight/tracking tables, padding values, shadow stacks, breakpoints

Both live in the project root. /minoan-frontend-design consumes both — creative direction from the design context, materials from the design tokens. Tokens are materials, not recipes: creative direction still leads.

Diagrams

Component hierarchies and token relationships can be visualized as Mermaid diagrams using beautiful-mermaid ASCII mode, then included directly in DESIGN.md files:

printf 'graph TD\n  Tokens --> Components --> Patterns' | \
  node ~/.claude/skills/beautiful-mermaid/scripts/mermaid.mjs -

For SVG export (e.g., in companion HTML docs), use -f svg -t github-light.

When Not to Use This Skill

  • For pattern research (how do other design systems implement tabs?), use /component-gallery
  • For creative direction and UI implementation, use /minoan-frontend-design
  • For pre-code design briefs, use /shape
  • For visual effects (shaders, WebGL), use /grainient or /rocaille-shader

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Search academic papers, build literature reviews, and synthesize research findings — combines Exa MCP (research_paper category, arxiv filtering) with arxiv-mcp-server for paper discovery, download, and deep analysis. Triggers on academic paper, literature review, research synthesis, arxiv, find papers, scholarly search.

日本語の概要は準備中です。原文の説明を表示しています。

tdimino/claude-code-minoan412026年9月28日 更新

Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.

日本語の概要は準備中です。原文の説明を表示しています。

tdimino/claude-code-minoan412026年9月28日 更新

This skill should be used when creating, auditing, or maintaining AGENTS.md files and Codex CLI configuration for any project—including initializing AGENTS.md for cross-agent compatibility (Codex, Cursor, Copilot, Devin, Jules, Amp, Gemini CLI), generating config.toml or .rules files, scaffolding .agents/skills/, converting CLAUDE.md to AGENTS.md, or auditing existing agent configs for bloat and staleness. Complementary to codex-orchestrator (which executes subagents; this skill creates the config files they consume).

日本語の概要は準備中です。原文の説明を表示しています。

tdimino/claude-code-minoan412026年9月28日 更新

Academic research skill for Biblical Hebrew, Semitic linguistics, cuneiform studies, and comparative Ancient Near Eastern research. Provides Sefaria API for Hebrew Bible, CDLI/ORACC for cuneiform databases, and web discovery via Omnisearch, Exa, Firecrawl, and Obscura for finding scholarly sources across JSTOR, Perseus, Persée, Google Scholar, and academia.edu. Triggers on Hebrew quotes, cuneiform, Sefaria, ANE research, Minoan, search for scholarship, find papers, literature review, scholarly search, academic search, Genesis/Tehom, Ugaritic, Talmudic sources, extract from PDF, OCR academic.

日本語の概要は準備中です。原文の説明を表示しています。

tdimino/claude-code-minoan412026年9月28日 更新

Build comprehensive ARCHITECTURE.md files following matklad's canonical guidelines — bird's-eye views, ASCII/Mermaid diagrams, codemaps, invariants, and layer boundaries. Triggers on document the architecture, create ARCHITECTURE.md, map this codebase, architectural overview.

日本語の概要は準備中です。原文の説明を表示しています。

tdimino/claude-code-minoan412026年9月28日 更新

Generate physically-based atmospheric scattering shaders — sky domes, planetary atmospheres, LUT-optimized pipelines, depth-aware post-processing. Four modes — sky-dome, atmosphere-post, planet, lut. Triggers on atmospheric scattering, sky shader, sunset rendering, planet atmosphere, Rayleigh scattering, Mie scattering, volumetric sky, sky dome, atmosphere post-processing, aerial perspective, transmittance LUT, sky rendering, realistic sky, planetary rendering.

日本語の概要は準備中です。原文の説明を表示しています。

tdimino/claude-code-minoan412026年9月28日 更新

tdimino のスキルをすべて見る

このスキルの問題を報告する