> ⚠️ 개발 런타임 전용 — 이 스킬은 MoAI-ADK(Claude Code) 환경을 전제한다. Claude Cowork(Desktop)에서는 .moai/config 의존으로 동작하지 않을 수 있다. Desktop 사용자는 moai-marketer:content-copywriting을 사용한다.

design-copywriting

Brand-aligned content-copywriting skill for marketing and product websites. Absorbed from the retired v2.x *-copywriting capability (per the content-copywriting absorption policy) at v3.2.0. Enforces anti-AI-slop rules, requires brand voice context, and outputs structured JSON per section.

Quick Reference

Entry Conditions

Before generating copy, verify all three conditions are met:

1. Brand voice context is loaded: .moai/project/brand/brand-voice.md exists or is provided inline.
2. Target page or section scope is explicitly stated (landing page, about, pricing, etc.).
3. Anti-AI-slop checklist is active (see below).

If brand-voice.md does not exist, stop and instruct the user to run the brand interview via /design (Path B brand-interview).

Output Format

All copy output is structured JSON with the following top-level sections:

{
  "page_type": "<landing|about|services|pricing|contact>",
  "sections": {
    "hero": { "primary": {...}, "variant_a": {...} },
    "problem": { "primary": {...}, "variant_a": {...} },
    "solution": { "primary": {...}, "variant_a": {...} },
    "features": { "primary": {...}, "variant_a": {...} },
    "cta": { "primary": {...}, "variant_a": {...} },
    "pricing": { "primary": {...}, "variant_a": {...} }
  },
  "metadata": {
    "tone_profile": "<string>",
    "word_count": <number>,
    "reading_level": "<string>"
  }
}

Each section must include at least one A/B variant (variant_a). The primary field is the recommended default.

Implementation Guide

Section Structure

hero: The first visible block. Subject must be the reader outcome, not the company name.

• headline: Max 12 words. Present tense. Outcome-first.
• subheadline: One sentence. Expands the headline mechanism. Max 25 words.
• cta_primary: Verb + noun. Max 5 words. ("Start your free trial", "See how it works")
• cta_secondary: Optional. Softer alternative. ("Learn more", "Watch demo")

problem: Acknowledges the reader's pain point before proposing a solution.

• headline: Frame the cost of the status quo. Specific, not abstract.
• body: 2-3 sentences. Use second-person ("you", "your team").

solution: Introduces your product or service as the answer.

• headline: Direct claim with mechanism. ("We automate X so you can Y.")
• body: 3-4 sentences. State what it is, how it works, and one concrete outcome.

features: Scannable list of capabilities.

• Each feature: title (max 5 words) + description (max 2 sentences) + optional metric.
• Maximum 6 features per section.

cta: Final conversion prompt.

• headline: Urgency or clarity without manipulation. Avoid "limited time" cliches.
• button_text: Same rules as hero cta_primary.
• supporting_text: Trust signal (money-back guarantee, no credit card, etc.)

pricing: Only include if explicitly requested in scope.

• headline: Framing statement. ("Simple pricing. No surprises.")
• Plans array: name, price, billing_period, highlights (max 5 bullet strings).

Anti-AI-Slop Checklist

Every piece of generated copy must pass this checklist before delivery:

Forbidden patterns — Reject any copy containing:

• Exclamation marks used more than once per section
• "In today's fast-paced world" or equivalent opening
• "Cutting-edge", "state-of-the-art", "revolutionary", "game-changing"
• "Leverage" as a verb (replace with "use", "apply", "deploy")
• "Seamless", "frictionless", "best-in-class" without quantified evidence
• Passive voice where active is grammatically equivalent
• Filler clauses: "at the end of the day", "the fact of the matter is"
• Generic benefits without concrete mechanism ("helps you succeed")
• Third-person self-reference in hero section ("Our company believes...")

Required patterns — Every section must have:

• At least one concrete number, percentage, or time reference
• Reader-centric framing (subject is "you" or reader outcome, not "we")
• Active voice for all capability claims
• Specificity: replace "grow your business" with a measurable outcome

Tone calibration — Load from brand-voice.md:

• Respect the tone spectrum defined (e.g., "playful" vs "authoritative")
• Match vocabulary preferences (avoid jargon if jargon_level: low)
• Preserve brand-specific terminology exactly as defined

A/B Variant Rules

Each section must include one alternative version (variant_a) that differs in:

• Headline framing angle (problem-first vs. solution-first)
• CTA verb choice
• Tone register (slightly more or less formal)

Do not generate more than two variants unless the scope explicitly requires it.

Brand Voice Integration

Load .moai/project/brand/brand-voice.md and apply:

1. tone: Overall register. If missing, default to "confident and direct".
2. vocabulary_preferences: Preferred and avoided terms. Enforce strictly.
3. audience_familiarity: Determines assumed knowledge level. Adjust jargon accordingly.
4. example_phrases: Use as stylistic anchors. Mimic sentence rhythm, not content.

If any _TBD_ markers are present in brand-voice.md, stop and request completion before proceeding.

Integration with design-brand-system

When both content-copywriting and design work are in scope (path B of /moai design):

1. Complete copy sections first and output structured JSON.
2. Pass the JSON copy output to design-brand-system as the content contract.
3. Design tokens and layout must accommodate copy length constraints (headline character counts, CTA button text length).

Advanced Patterns

Reading Level Targeting

Adjust copy complexity to the target audience's reading level:

• Technical audience (B2B SaaS developers): Grade 10-12. Allow domain terminology.
• Business audience (SMB owners): Grade 8-10. Avoid acronyms without expansion.
• Consumer audience (general public): Grade 6-8. Short sentences, concrete imagery.

Include the calculated reading level in the output metadata reading_level field using the Flesch-Kincaid grade level scale.

Microcopy Guidelines

Form labels, error messages, and UI strings follow stricter rules:

• Labels: noun phrases, 1-3 words, no punctuation
• Placeholders: example values, not instructions ("jane@company.com" not "Enter email")
• Error messages: what went wrong + how to fix it, no blame language
• Success messages: confirm the action, state the next step

Social Proof Integration

When testimonials or case study data are available in brand context:

• Quote exactly, do not paraphrase customer quotes
• Include company name, role, and metric (where permitted)
• Place after the problem section (reaffirms pain) or after features (reaffirms solution)

Anti-Slop Pattern Dictionary (absorbed from design-slop-check)

This dictionary is the canonical anti-AI-slop reference for the design pipeline. It was absorbed from the design-slop-check skill per docs/plugin-family-design/03-moai-design-processing.md §3.3 — content-copywriting is the canonical owner (proactive avoidance at generation time), while design-slop-check remains a downstream QA gate that references this dictionary. Reject or flag copy matching these patterns before delivery, and replace with concrete numbers / reader outcomes.

English clichés (Tier 1 — almost always slop)

• "Reimagine your X]"
• "Unleash your X]" / "Unleash the power of X]"
• "Empower your team to X]"
• "Transform the way you X]"
• "Supercharge your X]"
• "Revolutionize X]"
• "Next-generation X]"
• "Cutting-edge X]"
• "Powered by AI"
• "Built for the future of X]"
• "Where X] meets Y]" (marketing headlines especially)

English suspicious phrases (Tier 2 — context-dependent)

• "Game-changing", "Best-in-class"
• "Seamlessly", "Effortlessly", "Intuitively"
• "Leverage", "Synergy"
• "World-class", "State-of-the-art"
• "X], simplified" / "X], reimagined" / "X], unleashed"

Korean clichés (Tier 1 — 거의 항상 슬롭)

• "혁신적인 X]"
• "차세대 X]"
• "재정의하는 X]"
• "새로운 패러다임"
• "AI 기반의 X]"
• "한 차원 높은 X]"
• "지금까지 없던 X]"
• "당신의 X]를 변화시킬"
• "이제는 X]의 시대"

Korean suspicious phrases (Tier 2 — 문맥 확인 필요)

• "원활하게", "손쉽게", "직관적으로"
• "최고의", "최첨단의"
• "강력한", "막강한"
• "한 번에", "단숨에"
• "더 이상 X]에 시달리지 마세요"

Copy-structure anti-patterns

• A B C D adjective enumeration: "Faster. Smarter. Better. Simpler." — meaningless adjective list (형용사 나열)
• Colon-then-emphasis: "Headline]: The X that finally Y"
• Unquantified statistics: "Trusted by thousands of customers" (no concrete number)
• Needless negation → affirmation: "No more X]. Just Y]."

Downstream chaining note: for Korean output, an optional post-processing pass via a humanizer skill (e.g. moai-domain-humanize) may follow — this is an optional refinement, not a hard dependency.

Works Well With

• design-brand-system: Visual design must accommodate copy constraints
• design-iteration-loop: GAN loop evaluates copy quality in Design Quality and Completeness dimensions
• expert-frontend: Receives the JSON copy output for implementation
• 카피 정확도 셀프체크: 본 스킬의 brand-voice 정렬 체크리스트로 카피 정확도를 검증합니다. MoAI harness(moai)의 sync-auditor가 함께 설치된 환경에서는 해당 agent로 평가를 보강할 수 있습니다.

관련 스킬

| 단계 | 스킬 | 용도 | |---|---|---| | Post-검수 | moai-coworker:ai-slop-reviewer | AI 슬롭 검수 (필수) | | Post-검수 | moai-writer:korean-humanize | 한국어 AI 티 제거 (슬롭 검수 다음, 필수) |

Source: Absorbed from the retired v2.x *-copywriting capability v3.2.0 per the content-copywriting absorption policy. REQ coverage: (internal provenance omitted) Version: 0.1.0