landing — Claude Code skill

Build a landing page in four staged artifacts: value proposition → long-form copy → HTML → buyer-eyes review. The first three form the build pipeline; the fourth stress-tests the copy through the Decision Maker's eyes. Each step reads the previous artifact(s), so you can iterate on positioning, copy, visuals, and review — independently.

The copy stage follows April Dunford's Sales Pitch framework (Insight → Alternatives → Perfect World → Introduction → Differentiated Value → Proof → Ask) with positioning foundations from Obviously Awesome layered on top. The VP stage adds Jobs To Be Done (Christensen, 3 levels), an explicit USP Venn subtraction step against named direct competitors, and an enriched ICP (Sponsor ≠ Decision Maker + Top 3 Pains / Gains / Objections).

Install

Local (from this repo)

bash install.sh

Use --force to overwrite an existing installation without a prompt.

Remote (no clone)

bash <(curl -fsSL https://raw.githubusercontent.com/OWNER/landing/main/install.sh)

Both modes land at ~/.claude/skills/landing/. Restart Claude Code to pick up the skill.

Usage

Run the four subcommands from your project directory. The first three form the build pipeline (run in order); the fourth is a review you can run any time you want a sanity check on the copy:

Command	Input (read from CWD)	Output (written to CWD)
/landing vp	— (interactive interview)	value-proposition.md
/landing copy	value-proposition.md	copywriting.md
/landing html	copywriting.md	index.html
/landing review	value-proposition.md + copywriting.md	copy-review.md

/landing without arguments shows the menu.

What each subcommand does

• /landing vp — conducts a positioning interview: Dunford's 5 foundations (competitive alternatives, unique attributes, value, target customer characteristics, market category), Jobs To Be Done (functional + optional emotional/social), USP subtraction against named direct competitors, segments with enriched ICP (Sponsor vs Decision Maker, Top 3 Pains/Gains/Objections), Point of View, and public-voice guardrails (vendors not to mention, trigger event, recurring deliverables per tier, non-fit segments). Writes a structured value-proposition.md.
• /landing copy — reads value-proposition.md, walks you through any remaining discovery gaps (alternatives, insight, social proof, offer), then produces copywriting.md with named value buckets, 2–3 hero headline variants, primary + secondary CTA, and contextual objection handling.
• /landing html — asks for setup (conversion mechanism, brand colour, tone, logo, GTM ID), then assembles index.html from a pre-built HTML skeleton + section snippets. CSS and JS are inline. Only external dependencies: Font Awesome CDN and (optionally) the embedded conversion widget.
• /landing review — reads value-proposition.md + copywriting.md, adopts the Decision Maker of the primary segment as a single lens, and produces copy-review.md with three sections: What smells off (verbatim quotes + reaction), Why I'm not buying yet (3-5 objections, each anchored to a Pain / Gain / known Objection from the VP), What would convince me (gap analysis + targeted rewrites for the most toxic lines). If the VP pre-dates the enriched ICP fields, the prompt collects them inline before generating — no need to re-run /landing vp.

Language

• Chat replies are always in the user's language.
• Artifact files default to English. If you want the output in another language, tell the skill at any point ("rispondi in italiano", "artifact en español") and it will honour it.

Repo layout

skill/                      # copied to ~/.claude/skills/landing/ by install.sh
├── SKILL.md                # router + language rules
├── vp/prompt.md            # positioning architect (Dunford + JTBD + USP Venn + enriched ICP)
├── copy/                   # Dunford copywriter
│   ├── prompt.md
│   ├── questions.md        # discovery questions
│   └── template.md         # output structure
├── html/                   # assembler
│   ├── prompt.md
│   ├── mapping.md          # markdown → snippet mapping
│   ├── rules.md            # output rules + final checklist
│   ├── skeleton.html       # HTML shell (head, CSS, JS)
│   └── snippets/           # 12 section snippets (nav, hero, …)
└── review/                 # buyer-eyes copy review
    └── prompt.md           # Decision Maker lens, three-section critique

install.sh                  # local + remote installer
tests/                      # bash test suite (runs structure, content, install)
DEVPLAN.md                  # planned work, milestone-by-milestone

Tests

bash tests/test_all.sh    # runs all suites in order

Individual suites:

bash tests/test_structure.sh
bash tests/test_skill.sh
bash tests/test_vp.sh
bash tests/test_copy.sh
bash tests/test_review.sh
bash tests/test_html.sh
bash tests/test_install.sh