backend-repository

You can decouple business logic by abstracting data storage behind a repository interface, implementing interface-first design with ports, and returning typed domain entities. This centralizes data access rules and keeps your core logic portable.

Implementing user-scoped data access and soft-delete behavior requires repositories to filter queries by userId and exclude archived records by default. This enforces data isolation and preserves historical data without exposing deleted entities to the application layer.

Hydration and validation work by reading raw database records through a Zod schema typed to the domain entity before returning them. This validates data at the repository boundary, ensuring only correctly structured typed objects reach your business logic.

Yes, you can write portable queries across SQL and NoSQL backends by centralizing data access behind a repository interface. This abstraction decouples your business logic from specific database engines, enabling portable data layer implementations.

Centralizing error handling at the repository boundary ensures database exceptions are caught and translated before reaching business logic. This decouples your application from specific database error formats and provides consistent typed domain entity returns.

You configure backend repositories via constructor injection, passing dependencies directly into the class instead of reading environment variables inside. This pattern supports interface-first design, improves testability, and cleanly separates data access configuration from implementation.