Type Alias AutoBePhase

Union type representing the five phases of AutoBE's waterfall development pipeline.

This type defines the sequential phases that AutoBE follows to transform natural language requirements into a complete, production-ready backend application. Each phase builds upon the outputs of previous phases, following a structured waterfall model with internal spiral loops for error correction and quality assurance.

The Five Phases:

1. analyze: Requirements Analysis Phase

   • Transforms user conversations into structured requirements documents
   • Produces comprehensive markdown documentation (2,000-30,000+ characters)
   • Defines business model, functional requirements (EARS format), API guidance, ERD
   • Output: AutoBeAnalyzeHistory with requirements files

2. database: Database Design Phase

   • Designs and generates Prisma database schema from requirements
   • Organizes tables into domain-driven namespaces/components
   • Ensures referential integrity, normalization, and performance
   • Output: AutoBeDatabaseHistory with Prisma schema models

3. interface: API Specification Phase

   • Creates OpenAPI specification from requirements and database schema
   • Defines endpoints, operations, request/response schemas, authorization
   • Ensures type-safe API contract aligned with database
   • Output: AutoBeInterfaceHistory with OpenAPI document

4. test: E2E Test Generation Phase

   • Generates comprehensive E2E test suites for all API operations
   • Creates test scenarios, test functions, and authorization test setup
   • Ensures 100% operation coverage with realistic test data
   • Output: AutoBeTestHistory with test code files

5. realize: Implementation Phase

   • Generates actual NestJS controller and service implementation code
   • Creates collector functions (data fetching), transformer functions (data processing)
   • Implements authorization decorators, guards, and business logic
   • Output: AutoBeRealizeHistory with implementation code

Waterfall + Spiral Model:

While phases execute sequentially, each phase contains internal spiral loops where AI agents generate code, compilers validate it, and correction agents fix errors until 100% compilation success is achieved. This combines the structured progress of waterfall with the iterative quality assurance of spiral development.

State Machine Integration:

Each phase maintains a step counter that auto-increments when requirements change, triggering automatic invalidation and regeneration of downstream artifacts to maintain consistency across the entire pipeline.