Landfall
A CLI wizard that turns your landing page vision into structured, AI-ready prompts. Design the spec, then let your AI coding tool build it.
Ask AI about Landfall
Powered by Claude Β· Grounded in docs
I know everything about Landfall. Ask me about installation, configuration, usage, or troubleshooting.
0/500
Reviews
Documentation
Landfall
Design your landing page spec, then let AI build it.
Quick Start Β· Features Β· How It Works Β· MCP Integration Β· Contributing
8-step wizard β structured spec β AI-ready prompts
Landfall is a CLI tool that guides you through a structured wizard to fully spec out a landing page β style, tone, layout, sections, copy β then generates a sequence of focused prompts you feed to any AI coding tool. Instead of vague one-shot prompts that produce generic results, you get step-by-step instructions that reference a complete specification.
https://github.com/user-attachments/assets/aa97324d-e79e-4a5f-9e1b-939d1e9af618
Full demo (45s)
Quick Start
# Initialize Landfall in your project
npx landfall init
# Start the wizard
npx landfall dev
# Open http://localhost:3333 and walk through the 8 steps
When you're done designing, click Build on the final step. Copy the generated prompts into Claude Code, Cursor, Windsurf, or any AI coding tool β one prompt at a time β and watch your landing page come to life.
Features
8-step design wizard
- Style β Pick colors from 42 palettes, choose fonts from Google Fonts, set border radius and shadows, upload visual inspirations
- Tone β Define brand voice, target audience, do's and don'ts, example phrases
- Sitemap β Add and reorder pages with drag-and-drop
- Sections β Choose from 12 section types with 5 layout variants each (60 options total)
- Copy & Visuals β Write per-section copy instructions and attach reference images
- Navigation β Configure navbar style, footer columns, social links, newsletter signup
- Preview β Review wireframes across desktop, tablet, and mobile
- Build β Generate, review, and export the full prompt sequence
Prompt generation
- Produces a step-by-step build sequence: style system β layout β individual sections
- Each prompt is self-contained with all the context the AI needs
- Output targets Next.js + Tailwind CSS (the stack most AI tools handle best)
MCP server (optional)
- Exposes prompts as Model Context Protocol tools
- AI coding tools can pull the next prompt and mark steps complete automatically
- Real-time progress tracking in the Landfall UI
Everything is local
- All config stored as JSON files in a
landfall/directory - No accounts, no cloud, no telemetry
- Works offline after install
Step 1: Define your style β 42 color palettes, Google Fonts, visual keywords
Step 4: Choose sections β 12 types with 5 layout variants each
How It Works
1. npx landfall init Creates landfall/ with template JSON files
2. npx landfall dev Starts the wizard at localhost:3333
3. Walk through 8 steps Each change saves to JSON in real time
4. Click Build Generates prompts in landfall/prompts/
5. Feed to AI coding tool One prompt at a time, in order
The wizard writes everything to plain JSON files:
landfall/
βββ config.json # Project metadata
βββ style.json # Colors, fonts, keywords, shadows
βββ tone.json # Voice, audience, guidelines
βββ sitemap.json # Pages and structure
βββ navigation.json # Navbar and footer config
βββ pages/
β βββ home.json # Sections for each page
β βββ ...
βββ assets/ # Uploaded inspiration images
βββ prompts/
βββ build-sequence.json # Generated prompt sequence
You can edit these files directly or use the web UI β changes sync both ways.
MCP Integration
Landfall ships with an MCP server so AI tools can orchestrate the build automatically instead of manual copy-paste.
Add it to your .mcp.json or tool config:
{
"mcpServers": {
"landfall": {
"command": "npx",
"args": ["landfall-mcp"],
"env": {
"LANDFALL_PROJECT_PATH": "./landfall"
}
}
}
}
The MCP server exposes:
landfall_get_project_infoβ Project overviewlandfall_get_next_stepβ Next prompt to executelandfall_mark_step_completeβ Mark a step donelandfall_get_statusβ Build progress
Then tell your AI tool: "Build my Landfall project" and it handles the rest.
CLI Reference
| Command | Description |
|---|---|
landfall init | Scaffold the landfall/ directory with template configs |
landfall dev [--port 3333] | Start the wizard web app |
landfall build | Generate prompts without the UI |
Tech Stack
- CLI: Node.js, Commander.js, TypeScript
- Web app: Next.js 16, React 19, Tailwind CSS 4, Radix UI
- MCP:
@modelcontextprotocol/sdk - Generated output: targets Next.js + Tailwind CSS
Development
git clone https://github.com/AmElmo/landfall.git
cd landfall
npm install
npm run dev
The app runs at http://localhost:3000 in development mode.
To build the CLI separately:
npm run build:cli
Contributing
Contributions welcome! Feel free to open issues or submit pull requests.
If you're thinking about a larger change, open an issue first so we can discuss the approach.
License
MIT
