Unicorn Landing — Security skill for Claude Code

unicorn-landing — Claude Code Skill

A Claude Code skill that audits or builds landing pages against the 11 structural patterns shared by Stripe, Airbnb, Coinbase, DoorDash, Dropbox, GitLab, Instacart, and Gusto ($280B+ combined market cap).

What it does

**Audit mode** — give it a URL, code, or page description. Scores all 11 patterns PASS / PARTIAL / FAIL with one sentence of reasoning, then outputs a prioritized fix list ranked by conversion impact.

**Build mode** — describe your product. Generates a complete landing page spec: category-claim headline, hero copy under 30 words, free CTA, proof number, nav list, audience split recommendation, and footer structure.

The 11 Patterns

1. Category-defining headline — own THE category, not A category
2. Hero simplicity — ≤30 words total, one question answered ("am I in the right place?")
3. Two-sided marketplace split — both audiences addressable in 3 seconds (marketplaces only)
4. Product in the hero — real UI screenshot, not illustration or lifestyle photo
5. Social proof through scale — biggest true number, always
6. Minimal navigation — ≤6 items max
7. Personalization by segment — advanced; do this after patterns 1–6
8. Free CTA — enterprise sales happens after self-serve adoption
9. Page speed is invisible — LCP < 2.5s, CLS < 0.1
10. Footer as navigation hub — grows with product scope
11. Cross-page consistency — same fonts, colors, spacing everywhere

Installation

# Option 1 — clone directly into Claude skills directory
git clone https://github.com/peth-eth/unicorn-landing ~/.claude/skills/unicorn-landing

# Option 2 — clone anywhere and symlink
git clone https://github.com/peth-eth/unicorn-landing /path/to/skills/unicorn-landing
ln -s /path/to/skills/unicorn-landing ~/.claude/skills/unicorn-landing

Then invoke in any Claude Code session:

/unicorn-landing

Claude will ask for the page to audit or describe, then run through all 11 pa