Cursor + NestJS

What is the best Cursor workflow for NestJS backend development?
I’m looking for practical approaches for architecture, DTOs, services, repositories, tests, and maintaining consistency across a large codebase.

Hey, good request. For NestJS in a large codebase, the foundation of consistency is properly set up Project Rules .cursor/rules. Here are a few practical approaches:

Rules by architecture layers

  • Create separate rule files by area controllers, services, DTOs, repositories, tests with globs so a rule is only loaded when the agent works in matching files. This keeps context focused and prevents the prompt from getting too big in a large project.
  • Inside each rule, lock in conventions like module structure, DI pattern, where validation lives class-validator in DTOs, naming, error handling. The more specific examples you include in the rule, the more reliably the agent follows them across sessions.

Consistency across sessions

Workflow for features

  • For new features or unfamiliar parts of the codebase, use Plan mode. The agent first drafts a plan which modules, DTOs, services it will touch, you review it, then it implements. In large projects, this noticeably reduces drift from the architecture.
  • For test generation, keep a separate rule with an example of your test pattern Jest plus describe/it structure, mocks, fixtures. The agent will copy the style.

Context

  • Reference specific files and folders using @ instead of relying on the agent to find the right module on its own. In a large NestJS project, targeted context works better than guesswork.

Rules reference Rules | Cursor Docs

A similar breakdown of adapting Cursor in a team Java, but the ideas transfer Boosting Team Code Generation Adoption & Cursor UX: Our Enterprise Java Journey

If you tell me where consistency drifts DTO generation, services, or tests, I can give more targeted tips.