14 patterns · 33 working examples
Importing a customer's old data must be all-or-nothing AND safe to run twice — otherwise one failed import leaves half-moved data nobody can untangle.
14 patterns · 33 working examples
These aren't chapter titles — each one is an enforced rule your AI builds against. And the patterns don't just teach the shape: the working examples are real code that lands in your project.
Turning one address input into clean, queryable, geocoded columns
Storing third-party OAuth tokens so even a database breach can't expose them
The base shape behind every inbox, thread, and message a product needs
Adding a new messaging surface without standing up a new table for it
Tracking registrations, attendance, and submissions without a table each
Where reusable content — forms, curricula, templates — should actually live
Multi-stage sales pipelines with a real audit trail and forecast math
Generated invoices, line items, and payments that reconcile themselves
Clock-in and clock-out scheduling with break rules and clean payroll export
Becoming your own email sender without tanking your domain's reputation
Getting compliant US business texting past carrier filtering rules
Importing a customer's old system data without corrupting it on entry
Capturing signups safely before public registration is even open
A safe direct-to-storage upload path that skips routing files through your server
Bringing in a real customer's existing data becomes a one-shot script you can't repeat or audit.