Project Setup & Data Layer

This module lays the same foundation the original SKYWATCH course starts with: the monorepo that holds the frontend, the backend, and the types they share, plus the Postgres schema everything else reads from and writes to. The difference is how you get there. Instead of typing every file by hand, you write a prompt, read what the AI produces, and decide whether it’s actually right before it becomes part of your app.

Why This Matters

An AI coding assistant will happily scaffold a monorepo or write a Drizzle schema in seconds. It will just as happily get the workspace resolution wrong, skip a database-level invariant, or write a migration that looks plausible and is subtly broken. The gap between “the AI produced code” and “the AI produced correct code” is where this module lives. You need to know what correct looks like at this stage, or you have no way to tell the difference.

What You’ll Learn

• Prompting an AI coding assistant to scaffold an npm workspaces monorepo with three packages, and reviewing the result for the same buildless-shared-package setup the original course hand-wrote
• Catching a real class of mistake AI assistants make with workspace paths resolution, and the follow-up prompt that fixes it
• Prompting a Postgres schema into existence with Drizzle, then verifying the database-level invariants (a partial unique index, a singleton check constraint) actually landed, instead of trusting a confident-sounding summary
• The generate, review, and steer loop you’ll reuse in every lesson for the rest of this course

Outcome

By the end of this module, you’ll have the same working monorepo and migrated seven-table Postgres database as the original course, built by an AI assistant you directed and checked at every step, not one you trusted blindly.

Let’s get started.