Cohort de Marketing Skills

Skill by ara.so — Marketing Skills collection.

Marketing research and offer creation toolkit built by Academia Lendária. Provides 6 Claude Code skills for avatar research, competitor intelligence, trend hunting, swipe file organization, and offer creation. Clone-and-run system that loads skills automatically into Claude Code.

Installation

git clone https://github.com/marketingLendario/cohort-de-marketing.git
cd cohort-de-marketing

Copy environment template for optional API keys:

cp .env.example .env

Open in Claude Code:

claude

Skills in .claude/skills/ load automatically. Type / to see available skills.

Core Philosophy

Research → Offer → Copy → Ads (never skip this order)

Pesquisa antes da oferta (research before offer)
Voz do cliente sempre verbatim (customer voice always verbatim)
Mark assumptions as [SUPOSIÇÃO] when no real data exists
Offerbook approval before any copy/LP/ads

The 6 Skills (Aula 01)

`/avatar-funil`

Avatar research across 7 dimensions + synthetic focus group. Outputs MD + HTML + PDF.

Usage in Claude Code:

/avatar-funil

Produto: Curso de copywriting para coaches
Mercado: Coaches de vida e carreira Brasil
Público: Mulheres 30-45 anos com negócio online

Dimensions covered:

Demographics
Psychographics
Pain points & desires
Current solutions & frustrations
Buying triggers
Objections
Language patterns (verbatim phrases)

Output: avatar-[produto].md, avatar-[produto].html, avatar-[produto].pdf

`/espiao-do-concorrente`

Multi-source competitor intelligence dossier.

Usage:

/espiao-do-concorrente

Concorrente: [Nome ou URL]
Fontes: Meta Ad Library, Google Ads, Instagram, site, reviews

Sources analyzed:

Meta Ad Library (active ads)
Google Ads (search ads)
Social media presence
Website copy & structure
Customer reviews
Pricing & positioning

Output: espiao-[concorrente].md with structured intel

`/trend-hunting`

Identify emerging trends before saturation across 4 sources.

Usage:

/trend-hunting

Nicho: Marketing digital para psicólogos
Período: últimos 3 meses

Sources:

Google Trends
Reddit discussions
Twitter/X conversations
Industry publications

Output: trends-[nicho]-[data].md with trend score and timing

`/swipe-file`

Organize winning creatives by type/format/pattern.

Usage:

/swipe-file

Tipo: Ads do Facebook
Nicho: Infoprodutos
Padrão: Problem-Agitate-Solve

Categories:

Headlines
Hooks (video/carousel)
Body copy patterns
CTA formulas
Visual patterns

Output: swipe-[tipo]-[data].md with categorized examples

`/offerbook`

The Offer Book in 7 blocks. Outputs MD + DOCX using official template.

Usage:

/offerbook

Produto: [nome do produto]
Avatar: [referência ao avatar-funil gerado]
Concorrentes: [referência ao espiao gerado]

7 Blocks:

Contexto - Market positioning
Avatar - Who buys (from avatar research)
Problema - Pain landscape
Solução - Your mechanism
Oferta - What they get (stack)
Garantia - Risk reversal
Preço - Pricing strategy & logic

Output:

offerbook-[produto].md
offerbook-[produto].docx (uses templates/Template-Offerbook.docx)

`/design-md`

Generate DESIGN.md with brand identity (logo, colors, fonts). Used by /avatar-funil and /offerbook to render HTML with your branding.

Usage:

/design-md

Brand: Nome da Marca
Cores primárias: #HEX1, #HEX2
Fonte principal: Font Name
Logo: [URL ou caminho]

Optional: If not run, outputs default to Academia Lendária branding.

Output: DESIGN.md at project root

Configuration

Optional API keys in .env:

# Meta/Facebook (for Ad Library access)
META_ACCESS_TOKEN=your_token_here

# Google Ads (for competitor ad intel)
GOOGLE_ADS_DEVELOPER_TOKEN=your_token_here
GOOGLE_ADS_CLIENT_ID=your_client_id
GOOGLE_ADS_CLIENT_SECRET=your_secret

# OpenAI (for enhanced synthesis)
OPENAI_API_KEY=your_key_here

# Serper (for search/trend data)
SERPER_API_KEY=your_key_here

All keys are optional. Skills work in manual mode without keys (you provide data manually).

Typical Workflow

1. Research Phase

# Start with avatar research
/avatar-funil

# Then competitor intel (run for 2-3 competitors)
/espiao-do-concorrente

# Check trends
/trend-hunting

# Gather swipes
/swipe-file

2. Offer Creation Phase

# Optional: Set your brand
/design-md

# Create the offer book
/offerbook

3. Approval & Handoff

Review offerbook-[produto].docx with stakeholder. No LP, email, or ads until offerbook is approved.

Project Structure

cohort-de-marketing/
├── .claude/
│   └── skills/              # 6 skills (Aula 01)
│       ├── avatar-funil.md
│       ├── espiao-do-concorrente.md
│       ├── trend-hunting.md
│       ├── swipe-file.md
│       ├── offerbook.md
│       └── design-md.md
├── aula-01/
│   ├── GUIA-DO-ALUNO.html   # Start here (visual guide)
│   ├── docs/
│   │   ├── SKILLS-INDEX.md  # Name aliases (old → canonical)
│   │   ├── workflow.md
│   │   └── handoff.md
│   └── templates/
│       └── Template-Offerbook.docx
├── .env.example
└── README.md

Real Code Examples

Calling a skill programmatically (if building automation)

Skills are markdown-based and designed for Claude Code's slash command interface, but can be invoked programmatically:

// Example: Parse avatar research output
const fs = require('fs');
const matter = require('gray-matter');

function parseAvatarResearch(filepath) {
  const content = fs.readFileSync(filepath, 'utf8');
  const { data, content: markdown } = matter(content);
  
  // Extract verbatim quotes (customer voice)
  const quotes = markdown.match(/> "(.+?)"/g) || [];
  const cleanQuotes = quotes.map(q => q.replace(/> "/g, '').replace(/"/g, ''));
  
  return {
    meta: data,
    verbatimQuotes: cleanQuotes,
    rawContent: markdown
  };
}

const avatar = parseAvatarResearch('./avatar-curso-copy.md');
console.log(`Found ${avatar.verbatimQuotes.length} customer voice quotes`);

Integrating with Offerbook generation

// Example: Pass avatar data to offerbook
const avatarData = parseAvatarResearch('./avatar-produto.md');
const competitorData = JSON.parse(fs.readFileSync('./espiao-data.json', 'utf8'));

// Use in prompt for /offerbook
const offerPrompt = `
/offerbook

Produto: ${productName}
Avatar: ${avatarData.meta.targetAudience}
Pain points principais: ${avatarData.verbatimQuotes.slice(0, 3).join(', ')}
Concorrentes: ${competitorData.competitors.map(c => c.name).join(', ')}
`;

Common Patterns

Pattern: Research → Synthesis → Offer

# 1. Gather all research
/avatar-funil
/espiao-do-concorrente  # Run 2-3x for different competitors
/trend-hunting

# 2. Review outputs, highlight key insights

# 3. Create offer
/offerbook  # Reference the research files

Pattern: Verbatim Customer Voice

Always include literal quotes in avatar research. Mark assumptions:

## Linguagem do Avatar

### Verbatim (real quotes)
> "Eu não sei como cobrar sem parecer caro"
> "Meus posts não convertem, só dão like"

### [SUPOSIÇÃO] - Needs validation
- Provavelmente usa Instagram como canal principal

Pattern: Competitor Angle Gap

## Brechas de Ângulo

| Concorrente | Ângulo | Brecha |
|-------------|---------|--------|
| Concorrente A | "Venda todo dia" | Não fala de posicionamento |
| Concorrente B | "Copy que converte" | Ignora tráfego/audiência |
| **Nossa brecha** | "Posicionamento → Copy → Audiência" | Sistema completo |

Troubleshooting

Skills don't appear in Claude Code

# Check skills are in correct location
ls -la .claude/skills/

# Should see:
# avatar-funil.md
# espiao-do-concorrente.md
# trend-hunting.md
# swipe-file.md
# offerbook.md
# design-md.md

# Restart Claude Code

Old skill names (aliases)

Check aula-01/docs/SKILLS-INDEX.md for name mappings:

# Old name → Current name
/pesquisa-avatar → /avatar-funil
/analise-concorrente → /espiao-do-concorrente
/criar-oferta → /offerbook

DOCX generation fails

Ensure templates/Template-Offerbook.docx exists:

ls -la aula-01/templates/Template-Offerbook.docx

Missing API keys

Skills work without keys in manual mode. You provide data when prompted:

/espiao-do-concorrente

[Sem META_ACCESS_TOKEN configurado]
Por favor, cole manualmente os ads do Meta Ad Library para [Concorrente]:

Offerbook has no customer quotes

You skipped /avatar-funil. Research before offer (rule #1). Run avatar research first, then reference in offerbook.

Skill Name Reference

Canonical names (use these):

/avatar-funil - Avatar research
/espiao-do-concorrente - Competitor intel
/trend-hunting - Trend identification
/swipe-file - Creative organization
/offerbook - Offer book creation
/design-md - Brand identity setup

Support

Technical issues: Open issue at https://github.com/marketingLendario/cohort-de-marketing
Skill name confusion: Check aula-01/docs/SKILLS-INDEX.md
Content questions: Cohort community channel

Future Aulas

Aula 2: Funil e Páginas (funnel structure, landing pages, creative batches)
Aula 3: Tráfego e Criativos (traffic, ads, creative testing)
Aula 4: Dados e Receita (analytics, revenue optimization)

Skills publish week of each aula.

Built by Academia Lendária with Claude Code.

cohort-de-marketing-claude-skills

Cohort de Marketing Skills

Installation

Core Philosophy

The 6 Skills (Aula 01)

/avatar-funil

/espiao-do-concorrente

/trend-hunting

/swipe-file

/offerbook

/design-md

Configuration

Typical Workflow

1. Research Phase

2. Offer Creation Phase

3. Approval & Handoff

Project Structure

Real Code Examples

Calling a skill programmatically (if building automation)

Integrating with Offerbook generation

Common Patterns

Pattern: Research → Synthesis → Offer

Pattern: Verbatim Customer Voice

Pattern: Competitor Angle Gap

Troubleshooting

Skills don't appear in Claude Code

Old skill names (aliases)

DOCX generation fails

Missing API keys

Offerbook has no customer quotes

Skill Name Reference

Support

Future Aulas

`/avatar-funil`

`/espiao-do-concorrente`

`/trend-hunting`

`/swipe-file`

`/offerbook`

`/design-md`