Brand System Builder

Brand System Builder

Creates a comprehensive brand rules file that informs all other skills. Every skill that produces content should reference this file.

When to Use

Trigger phrases (exact match):

• "build the brand file"
• "create brand.md"
• "set brand rules"
• "define brand voice"
• "brand guidelines"
• "create brand system"
• "write brand rules"

Also triggers if:

• Starting a new pack without existing brand file
• repo-bootstrap reports no brand file exists

Not a trigger if:

• A brand file already exists and is valid
• User asks to edit specific content (not brand rules)
• User wants a style guide for web/print (this is for prompt workflows only)

Before Starting

Ask (or infer) these:

1. Brand name: What is this brand called?
2. Industry/Domain: What space does this brand operate in? (affects vocabulary)
3. Audience: Who are they talking to? (engineers, executives, consumers, etc.)
4. Tone direction: Pick one primary and one secondary
   • Primary: formal / casual / technical / playful / authoritative / friendly
   • Secondary: anything that complements
5. Existing brand materials: Does brand exist already? Logos, existing copy, website?
6. Output path: Where to save the brand file? (default: {repo}/BRAND.md)

Default if not specified:

• Brand name: derived from repo name
• Industry: technology/software
• Audience: professionals
• Tone: technical + direct
• Output: BRAND.md in repo root

Output Structure

Creates BRAND.md with these sections:

1. Brand Overview
2. Voice Dimensions
3. Vocabulary Rules
4. Typography (if applicable)
5. Color Palette (if applicable)
6. Content Examples
7. Anti-Examples
8. Channel Adaptations

Step-by-Step Process

Step 1: Gather Brand Intelligence

Ask the user for:

Minimum required:

• 3 links to content the brand has published (blog posts, website, ads)
• 1 example of content they loved and want to emulate
• 1 example of content that felt wrong for the brand
• One sentence: "This brand sounds like ___"

Nice to have:

• Competitor brands they admire
• Influencers/writers who nail the tone
• Anti-examples of tone they want to avoid

If user provides nothing: Use the default "technical + direct" persona and flag it as provisional.

Step 2: Define Voice Dimensions

Rate the brand on these 5 dimensions (1-5 scale):

Scoring rules:

• Always pick a number, never "depends"
• If unsure, default to 3
• Mark scores as "provisional" if based on limited samples

Translate scores into guidance:

Step 3: Define Vocabulary Rules

Positive list (use these words):

• List 10-15 words that capture the brand voice
• Examples: "precise", "fast", "reliable", "honest", "battle-tested"

Negative list (ban these words):

• List 15-20 overused corporate/AI phrases
• Examples: "cutting-edge", "seamless", "leverage", "robust", "holistic"
• See Appendix A for full ban list

Phrase patterns:

• What kind of sentences does this brand use?
• Short and punchy? Long and flowing? Mix?
• Active voice only? Or，偶尔被动？

Step 4: Define Visual Rules (if applicable)

If the brand system includes visual elements:

Color palette:

Primary:     #XXXXXX  (brand name)
Secondary:   #XXXXXX  (brand name)
Accent:      #XXXXXX  (brand name)
Neutral:     #XXXXXX  (brand name)
Background:  #XXXXXX  (brand name)
Text:        #XXXXXX  (brand name)

Typography:

Headlines:   {Font Name} — Bold
Body:        {Font Name} — Regular
Mono/Code:   {Font Name} — Regular

Spacing scale (8px grid):

• XS: 4px
• SM: 8px
• MD: 16px
• LG: 24px
• XL: 32px
• XXL: 48px

Step 5: Write Content Examples

Create 3 examples of brand voice in action:

Example 1 — Short form (one line) A product tagline, feature description, or headline.

Example 2 — Medium form (2-3 sentences) A feature explanation, value prop, or landing page blurb.

Example 3 — Long form (one paragraph) A full paragraph that shows how the voice holds up under pressure.

Rules for examples:

• Write real content, not placeholders
• If brand exists, derive from existing content
• If new brand, write aspirational content matching the dimensions
• Label each: "Strong brand voice ✓" or "Needs work ✗"

Step 6: Write Anti-Examples

Create 2-3 examples of content that violates the brand rules:

Format:

❌ WRONG: {bad example}
   Why it's wrong: {specific violation}
   Fix: {how to fix it}

✓ RIGHT: {corrected example}

Rules for anti-examples:

• Use real-sounding bad examples, not obvious placeholders
• Match the subject matter of the brand
• Make violations specific, not vague ("too corporate")

Step 7: Define Channel Adaptations

How does the voice adapt across formats?

Output Format

Create BRAND.md with this structure:

# {Brand Name} — Brand System

> One-line description of how this brand sounds.

**Provisional:** Mark as provisional if based on limited samples.

---

## Voice Dimensions

| Dimension | Score | What this means |
|-----------|-------|------------------|
| Formality | 3/5 | Professional but approachable |
| Optimism | 3/5 | Realistic, honest assessments |
| Playfulness | 2/5 | Serious, no humor in content |
| Competence | 4/5 | Assumes audience knows the domain |
| Novelty | 3/5 | Balanced between familiar and fresh |

---

## Vocabulary

### Use These Words
- word 1
- word 2
- word 3

### Never Use These
- overused phrase 1
- overused phrase 2

---

## Voice in Action

### Example 1 — Short

{content} ✓

### Example 2 — Medium

{content} ✓

### Example 3 — Long

{content} ✓

---

## Anti-Examples

❌ WRONG: {bad example}
   Why: {specific violation}
   ✓ RIGHT: {corrected version}

---

## Color & Typography

{If applicable — see structure above}

---

## Channel Adaptations

{Table of channel-specific rules}

Done Checklist

Before declaring complete:

• Voice dimensions scored (with provisional flag if needed)
• Vocabulary lists populated (use and ban lists)
• 3 real content examples that demonstrate the voice
• 2 anti-examples with specific violations and fixes
• Channel adaptations defined
• Color/typography sections added (if applicable)
• File saved to correct location (default: BRAND.md)
• File is valid markdown

Cross-Skill Orchestration

Called by:

• repo-bootstrap — if no brand file exists
• User directly — for brand refresh

Feeds into:

• prompt-author — all prompts reference brand voice
• bento-layout-director — visual brand rules inform layout
• lightrains-blog-writing (if used) — brand voice in content

Depends on:

• repo-bootstrap — must have a valid folder structure first

Edge Cases

Appendix A — Standard Ban List

These phrases are banned in almost every brand. Add to the negative vocabulary list:

• "cutting-edge", "best-in-class", "industry-leading"
• "seamless", "seamlessly", "effortless"
• "leverage", "utilize", "utilization"
• "robust", "comprehensive", "holistic"
• "robust", "comprehensive", "holistic"
• "delve", "delve into", "delve deeper"
• "landscape" (as buzzword)
• "navigate" (as metaphor)
• "foster", "facilitate", "ensure"
• "excited to announce", "thrilled to share"
• "game-changing", "revolutionary"
• "powerful", "ultimate", "perfect"
• "at its core", "fundamentally"
• "In this article, we will explore..."
• "It's worth noting that..."
• "Needless to say..."
• "That being said..."
• "Without further ado..."

Add industry-specific terms as needed.