nop-nodejs-backend

Use a fixed layering of Controller, Service, and PrismaService with modules under src/modules. Define models in schema.prisma with SQLite as the provider, and run prisma db push during development to sync tables without migrations.

Use the @map attribute on each model field and @@map on the model to set snake_case column and table names. Code then accesses properties like user.fullName while the database stores full_name, with Prisma handling the mapping automatically.

POST-only endpoints with all parameters in the JSON body reduce ambiguity for AI-generated code and match the existing product's interface style. This convention also simplifies later migration analysis since request shapes stay uniform across all operations.

Yes, change the datasource provider and connection string in schema.prisma while keeping the @map field mappings. The snake_case column naming already follows production database conventions, and UUID string keys remain stable at the API layer.

It explicitly targets demos, not production: no high-concurrency handling, complex transactions, internationalization, or multi-tenant support. It also avoids extra abstraction layers like repositories or domain services to keep business logic easy for AI to read and migrate.