菜单
Extract real brands (Wayback for rebuilds). Copy system, headline/CTA rules, trust surfaces, legal pages, SEO+structured data, anti-AI-slop, microcopy, DESIGN.md, W3C DTCG tokens, pSEO 5 types, GEO/AI search.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | brand-and-content-system |
| description | Extract real brands (Wayback for rebuilds). Copy system, headline/CTA rules, trust surfaces, legal pages, SEO+structured data, anti-AI-slop, microcopy, DESIGN.md, W3C DTCG tokens, pSEO 5 types, GEO/AI search. |
| metadata | {"version":"2.1.0","updated":"2026-05-03","effort":"medium","model":"haiku"} |
| license | Rutgers |
| compatibility | {"claude-code":">=2.0.0","agentskills":">=1.0.0"} |
| submodules | ["ai-search-geo.md","build-breaking-rules.md","documentation-and-codebase-hygiene.md","email-templates.md","grammar-audit.md","per-route-metadata.md","seo-and-keywords.md","social-automation.md"] |
| priority | 3 |
| pack | content |
| stage | stable |
| triggers | ["brand","content","copy"] |
| paths | ["org:website_build","org:website_build"] |
Extract real brand assets and apply the full copy system (anti-slop, SEO, trust surfaces, GEO) to every website build surface.
Model migration note (pass-77, 2026-06-09):
DALL-E→ GPT Image 1.5 +GPT-4o→ GPT Image 2 vision. Perplatform.openai.com/docs/deprecations. Brand extraction protocol unchanged.
Screenshot existing (Wayback if down). Extract logo/colors/fonts/tone. Never discard equity.
Color extraction (NON-NEGOTIABLE) — Screenshot with Playwright, GPT Image 2 vision extracts hex (logo priority), cross-ref logo, build palette, validate WCAG AA. NEVER invent, NEVER use Emdash defaults for clients, NEVER infer from category.
Second-pass verification (BUILD-BREAKING — pre-deploy) — After GPT Image 2 vision returns {primary, secondary, accent}:
min_hue_distance > 30° AND saturation>0.2, FAIL w/ diagnosticprimary derived from onevalidate-color-from-logo.mjs in build_validators.ts between brand-research and template-pickLogo-luminance + source-theme drives theme (NON-NEGOTIABLE) — Two-signal:
body background)detail:low) → match source theme verbatimLogo-vs-container contrast (BUILD-BREAKING — every render) — Every logo render (header, footer, hero, modal, splash, mobile menu, sidebar) MUST contrast container bg by ≥4.5:1 on logo's dominant chroma (not transparent pixels)
brand-mark-light.svg for dark bg, brand-mark-dark.svg for light bg) + CSS <picture><source media> swapsmagick -channel RGB -negate; color logos w/ text → GPT Image 1.5 w/ "same logo on transparent bg w/ text inverted to ")validate-logo-contrast.mjs — Claude Sonnet 4.6 vision samples logo bbox + container computed bg at 6bp + pixel sampling, fails if <4.5:1Brand-element extraction (logo is gold mine — extract DNA) — GPT Image 2 vision returns {font_family_guess, suggested_heading_font, suggested_body_font, font_weight, letterspacing, has_icon_mark, icon_mark_description, icon_mark_dominant_color, decorative_motif_description, motif_extractable (bool)}
--font-heading site-widemotif_extractable=true → crop icon-only region (magick logo.png -alpha extract -trim +repage), upscale 2-4× via Real-ESRGAN / GPT Image 1.5 variation, save as assets/brand-splash.png + assets/brand-mark.pngLogo singularity (BUILD-BREAKING — exactly ONE logo file per container) — Every logo container renders EXACTLY ONE logo asset
<img> tags; never render apple-touch-icon.png next to logo.svg in same containermagick logo-mark.png logo-wordmark.png +append, NOT at render timevalidate-logo-singularity.mjs parses dist/ HTML; count>1 descendant <img> or inline <svg> in same logo container = FAILLogo (NON-NEGOTIABLE) — Every project needs premium logo. See Skill 12 for full process.
Audit — logo found + rated ≥7/10 + works 16-512px, colors EXTRACTED, palette WCAG AA, typography + tone + messages identified.
https://maps.googleapis.com/maps/api/streetview?size=1200x800&location={lat},{lng}&source=outdoortypes: ["exterior", "storefront"]"Extract brand identity from this business storefront photo. Return JSON: {sign_text, sign_font_style, primary_color (hex), secondary_color (hex), accent_color (hex), logo_description, overall_aesthetic, confidence (0-1)}"color_source.#00E5FF, #50AAE3, #060610. Sora / Space Grotesk / JetBrains Mono.Banned copy words: delve | leverage | unleash | revolutionize | best-in-class | cutting-edge | discover | innovative | seamless | robust | synergy | elevate | empower | transformative
Banned patterns: "Welcome to" | "Discover [product]" | vague aspirational headlines | hedging ("may help you," "can potentially") | generic superlatives
Banned design tells: Inter as sole font | purple-blue gradients | uniform 16px border-radius everywhere | centered everything | Hero/Lucide as sole icon set | abstract 3D blobs | uniform fade-in on all elements | plastic AI stock photos
Fix: ask "Would the founder actually say this?" No → rewrite.
tokens.json → CSS vars + Tailwind config./integrations/{tool}) | Comparison (/compare/{a}-vs-{b}) | Use-case (/for/{audience}) | Template (/templates/{type}) | Location (/{city}-{service})sameAs + dated revision + ownershipSupreme policy layer governing all Claude Code behavior. Autonomy, one-line prompt interpretation, speed standards, emphasis signal processing, cross-skill coordination, done definitions, conflict resolution. Loaded every prompt.
Establish project thesis before first code. Infer product type from domain/folder/README. Identify users, business model, pSEO strategy, AI-native dev approach. Maintain PROJECT_BRIEF.md as source of truth.
Deep web research, competitor scanning, technology evaluation, and implementation planning. Decomposes work into vertical slices, identifies parallel workstreams, tracks assumptions with confidence levels, and designs the critical path for minimum wall-clock time.
Captures and evolves user preferences with confidence levels. Maintains Voice of the Customer model with exact language, dissatisfaction and aspiration signals. Handles promotion/demotion, global vs project scoping, auto memory system, and Omi wearable data integration.
Cloudflare-first platform selection. Decision trees for Workers, D1, R2, KV, DO, Queues, Vectorize, Containers, Sandboxes, Flagship, Agent Memory, Workflows v2. Default stack, override conditions, auth, data patterns, reliability.
Implements features in vertical slices, always starting with homepage. Enforces anti-placeholder rules — no lorem ipsum, no TODO stubs, no gray boxes. Real content, real images, real interactions. TypeScript strict mode, Zod validation, and structured file organization.