brand-to-design-md

Brand To DESIGN.md

Overview

Use this skill to turn public brand evidence into a token-ready DESIGN.md that a coding agent can use to generate consistent UI. When useful, also create a demo HTML page and verify it in a browser.

Start from the Google Labs DESIGN.md standard, not from example directories or third-party showcase pages. The output should be practical, source-aware, and executable by an agent.

Workflow

1. Confirm Scope And Standard

Clarify only if needed:

• Target brand URL or brand name.
• Required outputs: DESIGN.md only, or DESIGN.md plus demo HTML and screenshots.
• Target context: website, app UI, developer docs, product page, editorial asset, or general system.

Use the Google Labs DESIGN.md format as the structural reference: metadata, design tokens, component guidance, imagery rules, logo boundaries, and Do/Don't guidance.

2. Gather Evidence

Prefer primary and high-signal sources:

• Official homepage and product pages.
• Official brand, press, newsroom, media kit, or legal trademark pages.
• Public CSS, font declarations, SVG logos, favicons, images, and screenshots.
• Design agency case studies when they explain the identity system.
• Third-party brand databases only as secondary support.

Record evidence in four groups:

• Direct evidence: observed CSS, SVG fills, image colors, font names, layout behavior.
• Official prose: brand usage rules, typography descriptions, press-kit notes.
• Design rationale: motifs, concepts, color names, type-system explanations.
• Inference: practical agent decisions not explicitly stated by sources.

If access is blocked by a CAPTCHA, login, or security challenge, do not bypass it. Use accessible official pages, cached/search snippets, press kits, downloaded public assets, or clearly mark the limitation.

3. Extract Tokens

Translate evidence into roles, not just values.

Core token groups:

• colors: primary, secondary, tertiary, neutral, surface, on-primary, text, border, status/accent colors.
• typography: display, h1, h2, body, label, mono/code when present.
• rounded: small, medium, large, special brand shapes.
• spacing: base scale, common section gaps.
• components: buttons, cards, inputs, links, code panels, notices, navigation.
• imagery: photography style, illustration style, texture, gradient, motion, icon rules.

For each major token, prefer a role explanation:

• Where it came from.
• What it should be used for.
• What should not use it.

4. Compile DESIGN.md

Write a concise but usable file:

---
version: alpha
name: Brand Name
description: One sentence identity summary.
sources:
  - https://...
notes:
  - Evidence and limitation notes.
colors:
  primary: "#..."
typography:
  display:
    fontFamily: ...
components:
  button-primary:
    backgroundColor: "{colors.primary}"
---

## Overview
...

## Colors
...

## Typography
...

## Layout
...

## Components
...

## Imagery
...

## Logo And Usage
...

## Do's and Don'ts
...

Keep the writing concrete. Avoid generic claims such as "modern, clean, innovative" unless the evidence supports them and they are translated into specific design actions.

5. Generate A Demo When Useful

If asked for a demo, build a single-file HTML artifact in the workspace.

The demo should prove that the DESIGN.md is usable:

• First screen should be an actual branded product or site experience, not an explanation page.
• Include palette, typography, buttons, cards, forms, data/code panels, and one brand-specific motif.
• Use extracted public assets when allowed and useful.
• Recreate the brand's behavior and hierarchy, not only its colors.
• Keep the page responsive and readable on mobile.

For browser-based artifacts, use real browser or preview-environment verification:

• Open the file in an available browser or preview environment.
• Check console errors.
• Check horizontal overflow.
• Capture desktop and mobile screenshots.
• Confirm critical images and fonts fall back cleanly.

6. Report Output

Return:

• DESIGN.md path.
• Demo HTML path if created.
• Screenshot paths if created.
• Source summary.
• Verification result.
• Any limitations or inferred choices.

Quality Bar

A good output lets a future coding agent produce a page that feels recognizably tied to the brand without needing more context.

Check for:

• Tokens map to meaningful roles.
• The brand has distinct visual behavior beyond color.
• Logo and trademark boundaries are respected.
• Demo page uses the system in several UI contexts.
• The output distinguishes evidence from inference.

When To Read More

For a compact checklist and output templates, read references/workflow-checklist.md.