Database Designer

Infer schema from requirements. User never writes SQL.

Process

1. Analyze requirements

   • "Users can save expenses" → users, expenses tables
   • "Track categories" → categories table
   • "Monthly reports" → consider aggregation

2. Design schema

   • Tables and columns
   • Relationships (1:1, 1:N, N:M)
   • Indexes for performance

3. Generate migration

   • Create migration file
   • Apply to database
   • Update ORM models

Schema Patterns

Template-Specific

Supabase (nextjs-supabase)

-- Auto-generated, user doesn't see
create table expenses (
  id uuid primary key default gen_random_uuid(),
  user_id uuid references users(id),
  amount decimal not null,
  category text,
  created_at timestamptz default now()
);

PostgreSQL (fastapi-postgres)

# Alembic migration auto-generated
class Expense(Base):
    id = Column(UUID, primary_key=True)
    user_id = Column(UUID, ForeignKey('users.id'))
    amount = Column(Numeric, nullable=False)

Drizzle (hono-drizzle)

// Schema auto-generated
export const expenses = pgTable('expenses', {
  id: uuid('id').primaryKey().defaultRandom(),
  userId: uuid('user_id').references(() => users.id),
  amount: numeric('amount').notNull(),
});

User Experience

User: "I want to track expenses by category"

Internally:

1. Create expenses table
2. Create categories table
3. Add foreign key
4. Generate models
5. Create migration
6. Apply to database

User sees: "✅ Ready to save expenses"