跳到正文
cccskills
免费在 GitHub 公开

visual-brand-extractor

Extract visual branding (colors, typography, layout patterns) from a client's website and generate a style preset compatible with the HTML slides skill and a brand config JSON for the content asset creator. Uses WebFetch to read pages and analyzes CSS/HTML to identify the color palette, font pairings, and aesthetic patterns.

中文简介正在准备中。当前显示原文说明。

实测评分 未实测

查看安装方法

包含的文件(2)

  • SKILL.md15.1 KB
  • skill.meta.json264 B

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

Visual Brand Extractor

Extract a client's visual identity from their website and generate reusable style presets for slides and content assets. This is an agent-executed skill — the AI reads pages via WebFetch and performs the analysis directly.

Quick Start

Extract visual branding from https://vapi.ai for the Vapi client.

Inputs

InputRequiredDescription
Website URLYesClient's homepage or landing page URL
Client nameYesFor naming the output files
Additional pagesNoProduct page, docs page, etc. for richer extraction

Output

Two files saved to clients/<client-name>/brand/visual-identity.md:

  1. Slide preset — CSS custom properties, typography, and signature elements matching the format in skills/create-html-slides/STYLE_PRESETS.md
  2. Brand config JSON — Simple color/font config matching skills/content-asset-creator format

Process

Phase 1: Fetch Target Pages

Fetch 2-3 pages to get a representative sample of the brand:

  1. Homepage (mandatory) — the primary brand expression
  2. Product/feature page (if available) — deeper color and layout usage
  3. Blog or about page (optional) — secondary design context

Use WebFetch on each URL with a prompt like:

"Extract the full content of this page. I need: all color values (hex, rgb, hsl), font family names, CSS class names (especially Tailwind utility classes), any CSS custom properties/variables, meta tags, and the general structure of the page layout. Preserve exact color codes and font names."

Phase 2: Extract Color Palette

Analyze the fetched content to identify the color palette. Look for these sources in priority order:

2.1 CSS Custom Properties

Look for :root, html, or body blocks containing color variables:

--color-primary, --primary, --brand, --accent
--bg-*, --background-*
--text-*, --foreground-*

2.2 Meta Tags

Check for:

  • <meta name="theme-color" content="#..."> — often the primary brand color
  • <meta name="msapplication-TileColor" content="#...">

2.3 Explicit CSS Color Declarations

Search for color values in these properties:

  • background-color, background (including gradients)
  • color (text colors)
  • border-color, border
  • box-shadow (accent/glow colors)
  • fill, stroke (SVG logo colors)

2.4 Tailwind Utility Classes

If the site uses Tailwind CSS, map utility classes to hex values using the reference table below.

2.5 Classify Colors into Roles

RoleHow to Identify
bg-primarybackground-color on body, html, or outermost container. The most common background.
bg-secondaryBackground on cards, sections, or secondary containers. Slightly different from primary.
text-primarycolor on body or the most common text color.
text-secondaryMuted text variant — used on subtitles, descriptions, .text-gray-* elements.
accentMost prominent non-bg, non-text color. Found on buttons, links, CTAs, highlights. Also check <meta name="theme-color">.
accent-secondarySecond accent color, or gradient partner. Found on secondary buttons, hover states.
card-bgBackground of cards, modals, or elevated surfaces (if different from bg-primary).

2.6 Determine Theme Type

Check the luminance of bg-primary:

  • Dark theme: Dark background (black, navy, charcoal) with light text
  • Light theme: White or cream background with dark text
  • Mixed: Note if the site uses both (e.g., dark hero + light content sections)

Phase 3: Extract Typography

3.1 Find Font Sources

Look for font loading in this order:

  1. Google Fonts <link> tags — Font name is in the URL:

    fonts.googleapis.com/css2?family=Inter:wght@400;700
    

    → Font: Inter, weights: 400, 700

  2. Fontshare <link> tags — Similar URL pattern:

    api.fontshare.com/v2/css?f[]=clash-display@400,700
    

    → Font: Clash Display

  3. @font-face declarations — Extract the font-family value

  4. font-family CSS properties — Direct declarations on elements

3.2 Classify into Display vs Body

RoleHow to Identify
Display fontFont on h1, h2, .title, .heading, .hero-title, or the largest/boldest text.
Body fontFont on body, p, or general content elements.

If both use the same font family, it's a single-family pairing (like "Manrope + Manrope"). Note the different weights used for display (700-900) vs body (400-500).

3.3 Handle Proprietary Fonts

If the site uses a font NOT available on Google Fonts or Fontshare, map it to the closest available equivalent using the Font Fallback Table below.


Phase 4: Analyze Visual Patterns

Examine the CSS and page structure for these aesthetic signals:

SignalWhat to Look ForAesthetic Implication
Border radiusborder-radius values — large (12px+, 9999px) vs small (2-4px) vs noneRounded = friendly/modern. Sharp = corporate/bold.
Gradientslinear-gradient, radial-gradient in backgroundsGradient-forward = modern SaaS aesthetic
Shadowsbox-shadow frequency and intensityHeavy shadows = depth/elevation. None = flat/minimal.
SpacingPadding/margin sizes — generous (4rem+) vs tight (1rem)Airy = premium/minimal. Dense = information-rich.
Animations@keyframes, transition, transform presenceMotion-forward = energetic. Minimal = calm/professional.
Grid patternsBackground grid/dot patterns, decorative overlaysTechnical/developer aesthetic
Gradient orbsradial-gradient on positioned pseudo-elementsModern, atmospheric aesthetic
Borders as accentsColored border-left or border-top on sectionsEditorial, organized aesthetic

Synthesize the Vibe

Based on the signals above, compose:

  1. Vibe: 2-4 adjectives (e.g., "Clean, technical, developer-focused, modern")
  2. Layout description: 1 sentence (e.g., "Full-width dark sections with centered content and generous whitespace")
  3. Signature elements: 3-4 reproducible CSS patterns from the site

Phase 5: Generate Output

Create the file clients/<client-name>/brand/visual-identity.md with both output formats.

Output Template

# Visual Brand Identity: [Company Name]

**Extracted from:** [URL(s)]
**Date:** [YYYY-MM-DD]

---

## Slide Preset

**Vibe:** [2-4 adjectives]

**Layout:** [1 sentence describing dominant layout pattern]

**Typography:**
- Display: `[Font Name]` ([weight, e.g., 700/800])
- Body: `[Font Name]` ([weight, e.g., 400/500])

**Colors:**
```css
:root {
    --bg-primary: [hex];
    --bg-secondary: [hex];
    --text-primary: [hex];
    --text-secondary: [hex];
    --accent: [hex];
    --accent-secondary: [hex];
}
```

**Signature Elements:**
- [Element 1, e.g., "Subtle gradient orbs as background decoration"]
- [Element 2, e.g., "Rounded cards with 12px border-radius and light shadow"]
- [Element 3, e.g., "Accent-colored left border on feature blocks"]
- [Element 4, e.g., "Grid dot pattern overlay on hero sections"]

**Font Loading:**
```html
<link href="https://fonts.googleapis.com/css2?family=[Font1]:wght@[weights]&family=[Font2]:wght@[weights]&display=swap" rel="stylesheet">
```

---

## Brand Config (JSON)

For use with `content-asset-creator` and other skills:

```json
{
  "name": "[Company]",
  "primary_color": "[accent hex]",
  "secondary_color": "[accent-secondary hex]",
  "accent_color": "[accent hex]",
  "background": "[bg-primary hex]",
  "text_color": "[text-primary hex]",
  "font_heading": "[Display Font]",
  "font_body": "[Body Font]",
  "logo_url": "[if discovered, otherwise omit]"
}
```

---

## Extraction Notes

[Any notes about the extraction — proprietary fonts that were mapped, multiple themes detected, sparse CSS from JS-rendered sites, etc.]

Font Fallback Table

When a site uses a proprietary font not available on Google Fonts or Fontshare, use this mapping to find the closest available alternative:

Font CategoryProprietary ExamplesGoogle Fonts AlternativeFontshare Alternative
Geometric sansCircular, Roobert, Graphik, Product SansPlus Jakarta Sans, Outfit, ManropeGeneral Sans, Satoshi
Humanist sansProxima Nova, Calibri, Gill SansSource Sans 3, DM Sans, Work Sans—
Neo-grotesqueAkkurat, Aktiv Grotesk, Suisse Int'lSpace Grotesk, Archivo, Albert SansSwitzer
Modern serifTiempos, Canela, GT SectraFraunces, Cormorant, Playfair DisplayZodiak, Sentient
Slab serifSentinel, Clarendon, RockwellZilla Slab, Roboto Slab—
Classic serifMercury, Chronicle, MinionSource Serif 4, Lora, Libre BaskervilleBespoke Serif
MonospaceSF Mono, Berkeley Mono, Dank MonoJetBrains Mono, Fira Code, Space MonoJet Brains Mono
Display / decorativeGT Walsheim, Recoleta, CamptonSyne, Archivo Black, Bricolage GrotesqueClash Display, Cabinet Grotesk

Decision heuristic: If unsure which category a font falls into, look at its characteristics:

  • Round dots and geometric 'o' → Geometric sans
  • Varies stroke widths, warm feel → Humanist sans
  • Uniform strokes, neutral → Neo-grotesque
  • Serifs with high contrast → Modern serif

Tailwind CSS Color Reference

If the site uses Tailwind utility classes, map them to hex values:

Grays

ClassHexClassHex
gray-50#F9FAFBgray-500#6B7280
gray-100#F3F4F6gray-600#4B5563
gray-200#E5E7EBgray-700#374151
gray-300#D1D5DBgray-800#1F2937
gray-400#9CA3AFgray-900#111827
gray-950#030712

Slate (common for dark themes)

ClassHexClassHex
slate-50#F8FAFCslate-500#64748B
slate-100#F1F5F9slate-600#475569
slate-200#E2E8F0slate-700#334155
slate-300#CBD5E1slate-800#1E293B
slate-400#94A3B8slate-900#0F172A
slate-950#020617

Blues

ClassHexClassHex
blue-400#60A5FAblue-700#1D4ED8
blue-500#3B82F6blue-800#1E40AF
blue-600#2563EBblue-900#1E3A8A

Greens

ClassHexClassHex
green-400#4ADE80green-600#16A34A
green-500#22C55Egreen-700#15803D

Common Accent Colors

ClassHexClassHex
indigo-500#6366F1purple-500#A855F7
indigo-600#4F46E5violet-500#8B5CF6
emerald-500#10B981amber-500#F59E0B
teal-500#14B8A6rose-500#F43F5E
cyan-500#06B6D4orange-500#F97316

Note: Tailwind classes appear as bg-blue-600, text-gray-900, border-indigo-500, etc. Strip the property prefix (bg-, text-, border-) to get the color token, then look up the hex value above.

For custom Tailwind themes (e.g., bg-brand, text-brand-light), check the page's <script> tags for a tailwind.config object that defines custom colors.


Edge Cases

JS-Rendered Sites (Sparse CSS)

Many modern sites (Next.js, React) inject styles via JavaScript. WebFetch may return minimal inline CSS. In this case:

  • Focus on Tailwind utility classes in class attributes
  • Check <script> tags for embedded config objects
  • Look for <style> tags that contain CSS-in-JS output
  • If still sparse, ask the user: "The site appears to use dynamic styling. Could you share a screenshot or provide their brand guidelines directly?"

Multiple Color Schemes (Light/Dark Mode)

Some sites define both light and dark themes via prefers-color-scheme. Extract the default (non-media-query) theme. If a dark mode variant is detected, note it in the Extraction Notes section.

Too Many Colors

Enterprise sites may use dozens of colors. Focus on:

  • The hero/above-the-fold section for the primary palette
  • Button/CTA colors for accent identification
  • Frequency analysis — top 5-7 unique colors define the brand

No Usable Font Information

If font detection fails completely:

  1. Check if the site mentions font names in its CSS comments or config
  2. Look at the overall aesthetic and pick a matching pairing from the existing slide presets
  3. Default to a safe, distinctive pairing based on vibe: serif display + sans body for premium; geometric sans for modern tech

Example

Running this skill on https://vapi.ai might produce:

# Visual Brand Identity: Vapi

**Extracted from:** https://vapi.ai, https://docs.vapi.ai
**Date:** 2026-02-26

---

## Slide Preset

**Vibe:** Technical, developer-focused, clean, modern

**Layout:** Dark full-width sections with centered content, card-based feature grids, generous spacing

**Typography:**
- Display: `Inter` (700/800)
- Body: `Inter` (400/500)

**Colors:**
\```css
:root {
    --bg-primary: #0F172A;
    --bg-secondary: #1E293B;
    --text-primary: #F8FAFC;
    --text-secondary: #94A3B8;
    --accent: #22C55E;
    --accent-secondary: #3B82F6;
}
\```

**Signature Elements:**
- Subtle gradient orbs in background (green-to-blue radial gradients)
- Rounded cards with slate-800 background and subtle border
- Code-style monospace accents for technical terms
- Green accent on CTAs and interactive elements

**Font Loading:**
\```html
<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;700;800&display=swap" rel="stylesheet">
\```

(This is an illustrative example — actual extraction results will vary based on the live site.)


Tips

  • Homepage is usually enough. If the homepage has clear branding, you don't need additional pages. Fetch a second page only if the homepage is sparse or uses a different aesthetic from the rest of the site.
  • Buttons reveal the accent color. The most reliable way to identify the primary accent is to find the button/CTA background color.
  • Check the favicon/logo. SVG logos often contain the exact brand colors as fill values.
  • Don't over-extract. The goal is 5-7 colors, 1-2 fonts, and 3-4 signature elements. More than that creates noise, not signal.
  • When in doubt, simplify. A 3-color palette (bg + text + accent) with one font family at two weights makes a clean, usable preset.

Dependencies

  • Web fetch capability (for reading website pages)
  • No API keys or paid tools required

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Research supported advertising promises from a specific buyer's context, current product facts, customer language and observed ads. Turn them into static, video or copy ideas; match video ideas to actual recipes and save a product-scoped angle bank for write-video-ad-script. Product-led discovery can use verified facts without reviews. Reference persistence and reach are observations, not proof of performance.

中文简介正在准备中。当前显示原文说明。

gooseworks-ai/goose-skills12652026年10月24日 更新

Analyze ad campaign performance data (Google, Meta, LinkedIn) to identify what's working, what's wasting budget, and specific cut/scale/test recommendations. Runs statistical analysis, funnel diagnostics, and multi-channel budget reallocation with specific dollar-amount shift recommendations and scenario modeling.

中文简介正在准备中。当前显示原文说明。

gooseworks-ai/goose-skills12652026年10月24日 更新

For paid lead-gen and participant-recruitment ads, replaces vanity CPA with true CAC per qualified lead by joining ad-platform data with downstream funnel events, surfaces tracking gaps, and classifies every creative into Scale / Keep / Investigate / Cut.

中文简介正在准备中。当前显示原文说明。

gooseworks-ai/goose-skills12652026年10月24日 更新

The intake stage of a Meta ad harness. Researches the brand first, states what it assumes, and asks only what it cannot infer and what would change money, claims or legal exposure if guessed wrong — at most four questions a turn, in plain business language. Writes the brand's stable facts (ads/brand.md), the campaign brief, and a log of what was asked vs assumed. Use it when someone wants to run Meta ads and the brand's ad facts are not yet recorded.

中文简介正在准备中。当前显示原文说明。

gooseworks-ai/goose-skills12652026年10月24日 更新

Checks paid ads against the ad platform's rules (Meta, TikTok, Google, LinkedIn and others) and the brand's category rules: the copy, on-image or on-screen text, pictures and people. Rewrites what can be fixed honestly, adds the disclosures that are needed, and flags anything that needs a human or legal review, with one pass, fixed or flag line per ad and platform. Use alone when you already have the ad text or images and know where they will run; otherwise run the playbook for the job (e.g. make-static-ads, make-content, email-lifecycle).

中文简介正在准备中。当前显示原文说明。

gooseworks-ai/goose-skills12652026年10月24日 更新

Analyze the message match between your ads and landing pages. Checks if the promise in the ad copy carries through to the landing page headline, body, and CTA. Flags disconnects that kill conversion rates. Works with Google, Meta, and LinkedIn ads.

中文简介正在准备中。当前显示原文说明。

gooseworks-ai/goose-skills12652026年10月24日 更新

查看 gooseworks-ai 的全部技能

报告此技能的问题