# Marketing Brain: Claude Skill + Repo Template
## What this does
Turns one GitHub repo into your marketing team's shared AI context. Brand voice, ICP, strategy, funnel benchmarks, and campaign history live in plain markdown. Every Claude session, Cursor window, or agent reads the same source, so nobody re-explains the business to AI again.

## Setup (10 minutes)

1. Create a new private repo called `marketing-brain`.
2. Copy the folder structure below.
3. Paste the setup prompt into Claude (or run it in Claude Code inside the repo). It interviews you and fills every file.
4. Point your tools at it: open the repo in Claude Code or Cursor, or add the files to a Claude Project.

```
marketing-brain/
├── CLAUDE.md              # Rules every AI follows (read first)
├── brand/
│   ├── voice.md           # Tone, banned words, examples of good and bad copy
│   └── messaging.md       # Positioning, value props, proof points
├── audience/
│   ├── icp.md             # Firmographics, triggers, disqualifiers
│   └── personas.md        # Economic buyer, champion, influencer
├── strategy/
│   ├── plan.md            # This quarter's goals, bets, and budget
│   └── competitors.md     # Who we beat, who beats us, and why
├── funnel/
│   └── benchmarks.md      # Stage conversion rates, velocity, CAC, pipeline targets
└── campaigns/
    └── log.md             # What ran, results, what we learned
```

## The setup prompt

```
You are setting up a Marketing Brain: a repo of markdown files that every AI tool on my team will read as shared context. Interview me one section at a time, in this order: brand voice, messaging, ICP, personas, quarterly plan, competitors, funnel benchmarks, campaign log. Ask no more than 4 questions per section. After each section, write the finished markdown file with a clear H1, short bullet points, and real examples from my answers. Flag anything vague and ask me to sharpen it. Finish by writing CLAUDE.md: a 15-line rulebook that tells any AI to read brand/voice.md and audience/icp.md before writing anything, cite funnel/benchmarks.md for any number, and log new campaigns to campaigns/log.md.
```

## Keep it alive
- Update `funnel/benchmarks.md` monthly (or automate it with a scheduled GitHub Action that pulls from your BI tool).
- Add a line to `campaigns/log.md` after every launch. Results and lessons, three lines max.
- New hire onboarding = clone the repo and read CLAUDE.md.

---
Built by Clay Parker · clayparker-fullstackmarketer.vercel.app