backend-dev-guidelines

Layered architecture separates concerns into routes, controllers, services, and repositories. This pattern improves maintainability and testability by isolating business logic, data access, and HTTP handling into distinct layers that are easier to modify and test independently.

Use Zod for runtime type-safe validation of API inputs. Zod schemas define expected data structures, parse and validate incoming requests, and provide strong TypeScript inference, catching validation errors before they reach your business logic.

Integrate Sentry to capture and monitor errors across your microservices with rich context. Sentry tracks error frequency, performance metrics, and stack traces, enabling faster debugging and visibility into production issues.

Yes, BaseController provides a consistent response handling pattern across all endpoints. It standardizes HTTP status codes, response formatting, and error responses, reducing boilerplate and ensuring uniform API behavior.

Centralized configuration via unifiedConfig provides type-safe, validated access to environment variables with a single source of truth. This prevents runtime errors from missing or misconfigured environment values and improves code maintainability.

Prisma is optional but recommended for the repository layer. Any data access tool works with this architecture; Prisma integrates smoothly with the service-repository pattern and provides type safety for database queries.