# Corey Haines Launch Analysis — What Made It Viral

## The Numbers
- **14.9k GitHub stars** (and growing)
- **2.2k forks**
- Launched via a single tweet from `@coreyhainesco`
- Companion products: Conversion Factory (agency), Swipe Files (newsletter), Magister (AI agent), Coding for Marketers (guide)

## What He Did Right

### 1. Solved a Real, Immediate Problem
He didn't make "an AI tool." He made **marketing skills that technical founders can drop into Claude Code today**. The audience is specific: "technical marketers and founders who want AI coding agents to help with conversion optimization, copywriting, SEO, analytics, and growth engineering."

### 2. Massive Scope, Free
30+ skills covering the entire marketing stack — SEO, CRO, copywriting, cold email, pricing, launch strategy, analytics, ad creative, referral programs, churn prevention, and more. All open source. That scope is what creates the "holy shit" reaction.

### 3. Architecture That Shows Expertise
The ASCII art dependency diagram in the README is genius. It visually communicates:
- These skills are interconnected (not random files)
- There's a foundation layer (`product-marketing-context`)
- They're organized by marketing function
- They cross-reference each other

This signals "this was built by someone who deeply understands marketing" — not just prompt engineering.

### 4. Platform-Agnostic Positioning
"Works with Claude Code, OpenAI Codex, Cursor, Windsurf, and any agent that supports the Agent Skills spec." This dramatically expands the audience beyond Claude users.

### 5. Multiple Monetization Touchpoints (But Not Gated)
The skills are free. But the README links to:
- **Conversion Factory** — his paid agency
- **Swipe Files** — his newsletter/community
- **Magister** — his paid AI agent product
- **Coding for Marketers** — his companion guide

The open-source repo is a top-of-funnel magnet for his entire business ecosystem.

### 6. Contributions Welcome
"Found a way to improve a skill or have a new one to add? Open a PR." This creates community ownership and ongoing engagement.

### 7. Timing
Launched right as the "Claude Code skills" meme was peaking on Twitter. Thariq's Anthropic post, Karpathy's autoresearch, Ole Lehmann's thread — all in the same week. Corey surfed the wave.

## What We Can Learn

| Corey's Move | Our Equivalent |
|---|---|
| 30+ marketing skills | 3 deeply specialized visual research skills (+ supporting scripts) |
| Product-marketing-context as foundation | Brand config + research pipeline as foundation |
| ASCII dependency diagram | Pipeline flow diagram (DISCOVER → CAPTURE → EXTRACT → ANALYZE → PACKAGE → BUILD → QA) |
| Platform-agnostic | Target Claude Code + OpenClaw + any AgentSkills spec agent |
| Newsletter/agency as backend | CE portfolio / consulting as backend |
| Easy install | One-command install: `openclaw skills install visual-research` or git clone |
| Companion guide | "Visual Research Methodology" article/video showing real output |

## Key Viral Mechanics

1. **Show, don't tell** — The tweet should show a real output (a competition page we built) alongside the skills that produced it
2. **The "complete toolkit" angle** — Position as "everything you need to do X" not "a helper for X"
3. **Credibility signal** — Link to actual live pages built with these skills as proof
4. **Specificity > Generality** — "Visual competitive research skills" is more interesting than "research skills" because it's a defined workflow people can immediately picture
5. **Thread format** — Main tweet + breakdown thread showing each skill + what it does + example output