Data Models

The fastest way to get a useful app is to be specific about your data. Give CodeSky a list of entities and their fields, and it produces the whole vertical slice for each one — table, model, API, and UI.

What a data model produces

• A migration file (SQL or framework-specific).
• An ORM model (Sequelize, EF Core, SQLAlchemy — matched to your backend).
• CRUD endpoints under /api/{entity} with pagination and filtering.
• List, detail, create, and edit pages on the frontend, wired to those endpoints.

Example prompt

"Patients have name, dob, phone, and a list of appointments.
 Appointments belong to a patient and a doctor, with start_time and status."

From that you get two tables, foreign keys in both directions, list pages with search by name, and a "book appointment" form that picks an existing patient and doctor.

Describe relationships plainly

• "each order has many line items" → one-to-many with a foreign key.
• "a student belongs to one class" → a class_id on students.
• "products and tags are many-to-many" → a join table, generated for you.

Evolving the model safely

• Add a field by prompt — "add a `priority` enum to tasks (low/medium/high)" — and the migration, model, API, and form update together.
• Ask for a migration whenever the schema changes so production can be updated in order.
• Mention soft-deletes, timestamps, or an audit trail and they're wired through both the model and the queries.

Common mistakes to avoid

• Vague fields. "user info" gives you a guess; "name, email (unique), role" gives you the right columns.
• Forgetting relationships. Say how entities connect, or you'll get isolated tables with no foreign keys.
• Renaming in the UI only. Rename in the data model so the change flows through the whole stack.

FAQ