Get started

This is not a boilerplate

Boilerplates give you a folder of files and wish you luck. You spend weeks figuring out how everything connects, wiring auth to payments to email, and by the time it works you've already burned through the part where building should feel fun.

This is a build system. You describe what you want and AI agents build it. Not one general-purpose chatbot guessing at your codebase. A coordinated team of specialists that each own a specific layer: database, API, frontend, design, testing. They don't step on each other's work because each one knows exactly what files it touches.

The framework ships with auth, payments, email, storage, cron jobs, and a design system already wired together. That's your starting point, not your finish line.

How it works

You open Claude Code and follow five steps. Each one is a single command you type:

/discover    →  research your market, define your product
/setup       →  configure everything automatically
/mvp-spec    →  generate specs for every feature
/mvp-build   →  build them, one by one

You're not writing prompts from scratch or figuring out what to ask. The system knows what it needs and either finds the answer itself or asks you the right question at the right time.

What you actually get

A production codebase. Auth with email, magic links, and social login. Stripe payments with tiered pricing. Transactional email. File storage with security. Background jobs. Bot protection. All of it passing type checks and building cleanly before you touch it.

An AI build system. Specialist agents that plan, build, test, and ship features. A quality gate that catches problems before you see them. Pattern libraries that keep the code consistent as it grows.

A discovery pipeline. Research agents that go out and study your market, find the right tools, and validate demand before you write a line of code. By the time you start building, the hard decisions are already made.

A command library. Beyond the core pipeline, you get commands for security audits, penetration testing, performance analysis, design competitions, logo generation, and more. Each one is a focused workflow, not a generic prompt.

What's in these docs

The Pipeline section walks through the five core commands. What You Get covers every layer of the framework. The AI System explains how the agents work together. Commands documents everything you can run. Quality & Security covers the audit and testing tools.

You don't need to read all of this to start. Open Claude Code, type /discover, describe your idea, and follow the conversation. Come back here when you're curious about what's happening under the hood.