Sofka HTML Brand — Document Generator

When to Use

• Creating branded HTML deliverables for client presentations
• Converting markdown deliverables to branded HTML (pipeline mode)
• Upgrading existing HTML documents to Sofka Design System v5
• Batch processing multiple files to brand compliance
• Generating executive, technical, or transformation documents
• Building self-contained HTML reports with WCAG AA accessibility
• Rendering Mermaid diagrams with Sofka high-contrast theme
• Creating carousel slide-decks for client-facing executive proposals
• Creating presentation slides with arrow navigation for socialization sessions

When NOT to Use

• Multi-page web applications with routing → use a framework (React, Vue)
• Interactive dashboards with live data → build a dedicated app
• Print-only documents → use PDF generation tools
• Content writing → sofka-ux-writing for microcopy and readability
• Dark-theme HTML for web deployment → this skill generates light-content with dark hero/footer

Assumptions & Limits

• Output is single-file HTML with inline CSS; font <link> tags are the only external dependency
• Design System v5: orange #FF7E08 primary, Clash Grotesk display, Inter body
• Body background is #FFFFFF (white), NOT #EFEAE4 (crema) — learned from production contrast issues
• Does NOT handle multi-page apps, routing, or state management
• Does NOT embed base64 images (bloat); use relative paths or CDN URLs
• Maximum 15 sections per document; beyond that, split into separate deliverables
• Mermaid diagrams use CDN (cdn.jsdelivr.net/npm/mermaid) — only external JS dependency

Usage

/sofka-html-brand executive ./output/brief.html
/sofka-html-brand technical                       # outputs to current directory
/sofka-html-brand --batch ./legacy-docs/          # upgrade 3+ files in parallel
/sofka-html-brand --pipeline ./outputs/           # convert all .md files to branded HTML

Parse $1 as document type (executive, technical, transformation, --batch, --pipeline) or path. Parse $2 as output path.

Parameters:

• {MODO}: piloto-auto (default) | desatendido | supervisado | paso-a-paso
  • piloto-auto: Auto para generación rutinaria, HITL para decisiones de marca y accesibilidad.
  • desatendido: Cero interrupciones. Supuestos documentados.
  • supervisado: Autónomo con reportes en milestones. Preguntas solo en decisiones de marca.
  • paso-a-paso: Confirma antes de cada componente y decisión de diseño.
• {FORMATO}: html (default) | markdown | dual
• {VARIANTE}: ejecutiva (~40%) | técnica (full, default)

Before Generating

Load reference materials:

Read ${CLAUDE_SKILL_DIR}/references/design-tokens.md

For batch operations, markdown pipeline, or edge cases:

Read ${CLAUDE_SKILL_DIR}/references/operations-guide.md

Document Type Decision Tree

Is the primary audience C-level / board / stakeholders?
├─ YES → Is it a short, visual, self-navigable piece?
│   ├─ YES → CAROUSEL (8-10 horizontal slides, dot nav, swipe)
│   │   Goal: invite to action in 3 min
│   │   Slides: 8–10, visual-first, CTA final
│   │
│   └─ NO → EXECUTIVE
│       Goal: decision support in 15 min
│       Sections: 8–12, KPI-dense, lead with metrics
│
└─ NO → Is it for a live presentation / socialization session?
    ├─ YES → PRESENTATION SLIDES (20-25 full-screen, ← → nav)
    │   Goal: structured walkthrough in 45-60 min
    │   Slides: 20–25, speaker notes, progress bar, F for fullscreen
    │
    └─ NO → Is it about architecture, APIs, or technical decisions?
        ├─ YES → TECHNICAL DEEP-DIVE
        │   Goal: engineer/architect understanding
        │   Sections: 10–15, diagrams, ADRs, code
        │
        └─ NO → Multi-year roadmap or business transformation?
            ├─ YES → TRANSFORMATION DIGITAL
            │   Goal: rally business + tech
            │   Sections: 8–10, "why" first, timeline + ROI
            │
            └─ NO → Is it a markdown pipeline conversion?
                ├─ YES → PIPELINE MODE
                │   Goal: faithful branded rendering of existing .md
                │   Sections: auto-detected from h2 headers
                │
                └─ NO → Default to EXECUTIVE (safest for mixed audiences)

Document Structure

Every Sofka HTML deliverable follows this skeleton:

<!DOCTYPE html>
<html lang="es">
<head>
  <!-- charset, viewport, OG tags, fonts, Mermaid CDN, inline <style> -->
  <script>mermaid.initialize({...Sofka high-contrast theme...})</script>
</head>
<body>
  <a href="#main" class="skip-link">Ir al contenido principal</a>
  <!-- optional: internal-banner for INTERNAL docs -->
  <header class="hero">         <!-- black bg, orange bottom border -->
    <div class="hero-inner">
      <div class="hero-logo">sofka_</div>
      <div class="hero-meta-badges">...</div>
      <h1>Title <span>Highlight</span></h1>
      <p class="hero-subtitle">...</p>
      <div class="hero-kpis">...</div>  <!-- 3-4 KPIs -->
    </div>
  </header>
  <nav class="toc" aria-label="Navegación del documento">
    <div class="toc-inner">...</div>  <!-- sticky, horizontal scroll -->
  </nav>
  <main class="container" id="main">
    <!-- content: markdown-converted or hand-built sections -->
  </main>
  <footer class="site-footer">...</footer>
  <script>/* TOC tracking */</script>
</body>
</html>

Two Generation Modes

Mode 1: Hand-Built HTML (from scratch)

For new documents where you compose the HTML directly. Use numbered sections with class="section", component classes from design-tokens.md, and the full component library.

Mode 2: Markdown Pipeline (convert .md → branded HTML)

For converting existing markdown deliverables. This is the production-proven path used to generate 18+ deliverables at once.

Pipeline:

1. Parse markdown with npx marked --gfm
2. Apply bridge CSS (in design-tokens.md § "Markdown Bridge CSS") that maps bare HTML elements to DS v5 styles
3. Post-process: convert evidence tags [DOC] → <span class="badge badge-doc">DOC</span>
4. Post-process: convert mermaid code blocks to <pre class="mermaid"> elements
5. Extract h2 headers for TOC generation
6. Wrap in DS v5 shell (hero, TOC, footer) with per-deliverable metadata

Key insight: In pipeline mode, the markdown produces bare <h2>, <table>, <p>, etc. The bridge CSS maps these to DS v5 styles without requiring class names. This is what makes batch conversion possible.

Color Rules

Design System v5 uses yellow for success states because it maintains brand coherence with the warm Sofka palette — green introduces a cold tone that clashes.

Green (#42D36F), teal, violet, and pink exist only for charts and data visualization — never for semantic states.

Contrast Rules (Production-Learned)

ABSOLUTE PROHIBITION: Never white (#FFFFFF) text on crema (#EFEAE4) background. Contrast ratio is only 1.16:1 — invisible.

See references/design-tokens.md for the complete CSS variable system.

Mermaid Diagram Configuration

All Mermaid diagrams use the base theme with Sofka high-contrast variables. This ensures:

• Light-colored node fills (#FFF3E0, #FFFFFF, #F4F0EC) with black text
• Orange borders on primary nodes
• Gray borders on secondary/tertiary nodes
• White edge label backgrounds
• Inter font family

The complete Mermaid config is in references/design-tokens.md § "Mermaid Theme Configuration".

Evidence Badge System

Markdown evidence tags are converted to colored badges in HTML:

Content Density by Document Type

Component Usage by Document Type

Generation Workflow

Phase 1: Plan

1. Determine document type (decision tree above) or pipeline mode
2. List sections with IDs
3. Assign components per section using the type table
4. Draft hero metadata: title, phase badge, version, subtitle, 4 KPIs

Phase 2: Build

1. Load design-tokens.md for complete CSS
2. For pipeline mode: run markdown through npx marked --gfm, apply bridge CSS, post-process evidence tags and Mermaid blocks, extract TOC from h2 headers
3. For hand-built: compose sections with numbered headers and DS v5 components
4. Assemble: head (fonts, Mermaid init, CSS) → skip-link → hero → TOC → main → footer → JS

Phase 3: Quality Gate

1. Read top to bottom: any placeholder text remaining?
2. Contrast audit: ALL text on ALL backgrounds meets WCAG AA (4.5:1 body, 3:1 large)?
3. Color audit: only brand + semantic colors? No green for success?
4. Mermaid check: all nodes have light fills with black text?
5. File size check: under 500KB?

Anti-Patterns

Constraints

Trade-offs

Edge Cases

Example: Good vs Bad

Good hero section:

<header class="hero" style="background: var(--sofka-black); border-bottom: 8px solid var(--sofka-orange);">
  <div class="hero-inner">
    <div class="hero-logo">sofka_</div>
    <h1>Core Banking <span>Modernization</span></h1>
    <div class="hero-kpis"><!-- 4 KPIs --></div>
  </div>
</header>

Bad hero section:

<!-- WRONG: hardcoded colors, green for success, no brand font, 6 KPIs -->
<header style="background: #333; border: 1px solid gray;">
  <div style="font-family: Arial; color: white;">Sofka</div>
  <h1 style="color: #00ff00;">CORE BANKING MODERNIZATION</h1>
  <div><!-- 6 KPIs crammed together --></div>
</header>

Good Mermaid config:

mermaid.initialize({
  startOnLoad: true, theme: 'base',
  themeVariables: { primaryColor: '#FFF3E0', primaryTextColor: '#000000', ... }
});

Bad Mermaid config:

// WRONG: default theme creates dark fills with unreadable text
mermaid.initialize({ startOnLoad: true, theme: 'default' });

Validation Gate

Before delivering any HTML document, verify:

• Document type matches audience (executive/technical/transformation/pipeline)
• All colors use CSS variables from Design System v5 (no hardcoded hex outside tokens)
• Typography: Clash Grotesk for display, Inter for body (no substitutions)
• Hero has 3-4 KPIs maximum with orange highlight span
• Body background is #FFFFFF, NOT #EFEAE4
• All body text uses --sofka-gray-900 (#111110) — never white on light bg
• Cards/TOC use --sofka-gray-50 background, not --sofka-white
• Table cells use --sofka-gray-100 background with --sofka-gray-900 text
• Mermaid uses theme: 'base' with light fills and #000000 text
• Semantic states use correct colors (yellow=success, NOT green)
• WCAG AA contrast ratio met on all text (4.5:1 body, 3:1 large)
• Evidence badges converted from [TAG] to colored spans
• File size under 500KB
• Skip-link present: <a href="#main" class="skip-link">
• Single-file HTML with no external deps (except font + Mermaid CDN)
• lang="es" (or appropriate language) on <html> element
• No placeholder text remaining in output
• TOC links match h2 section IDs (max 8)
• Footer has Sofka tagline: "Construido por profesionales, potenciado por la red agéntica de Sofka."

Batch Processing

When upgrading 3+ files at once, use parallel sub-agents. Read references/operations-guide.md for the squad pattern, pipeline script approach, and error handling.

Reference Files

Agent Prompts

Cross-References

• sofka-ux-writing: UX writing standards for microcopy and readability
• sofka-design-system: Design system component library that HTML Brand implements
• sofka-executive-pitch: Executive-facing deliverables that use HTML Brand
• sofka-mermaid-diagramming: Mermaid diagram patterns (uses same high-contrast theme)

Output Format Protocol

Default output is self-contained HTML. This skill always produces HTML — it is the brand rendering engine.

Output Artifact

Primary: {NN}_{Deliverable}_{Client}.html — Brand-compliant HTML deliverable with Design System v5 tokens, WCAG AA accessibility, hero KPIs, sticky TOC, evidence badges, Mermaid diagrams.

Secondary: Brand audit report, contrast validation, component usage audit.

Design System: v5 "Dark Authority" | Last Updated: 2026-03-16

Autor: Javier Montaño | Última actualización: 16 de marzo de 2026