Skip to content

ARCHITECTURE.md

Latest commit

 

History

History
99 lines (61 loc) · 1.79 KB

ARCHITECTURE.md

File metadata and controls

99 lines (61 loc) · 1.79 KB

Architecture

start-it-cli is organized around a typed generation pipeline.

Main Flow

  1. src/cli.ts
  2. src/workflow.ts
  3. src/generator.ts
  4. stack-specific generator
  5. AI-control artifact injection

Pipeline Stages

1. CLI Input

src/cli.ts handles:

  • interactive prompts
  • command-line flags
  • partial prompt fallback
  • normalization into ProjectConfig

ProjectConfig is the contract between the CLI and generation logic.

2. Workflow Registry

src/workflow.ts defines:

  • available app types
  • available stacks per app type
  • framework inference per stack

Generation Paths

Backend

Backend generation is mostly builder-based.

  • src/templates/node.ts
  • src/templates/nest.ts
  • src/templates/fastapi.ts

Those return TemplateConfig objects that are written to disk by src/generator.ts.

Frontend

Frontend generation uses a baseline-plus-customization model.

  • src/frontend/scaffold.ts

Behavior:

  • prefer provider-native baseline when possible
  • fall back to local baseline when necessary
  • apply deterministic customizations afterward

AI / ML

AI/ML generation is deterministic and direct-to-filesystem.

  • src/aiml/scaffold.ts

DSA

DSA generation is deterministic and workspace-oriented.

  • src/dsa/scaffold.ts

AI Guidance Layer

Every generated project gets:

  • .cursorrules
  • .agentignore
  • docs/AGENTS.md
  • docs/instructions.md

Composition lives in:

  • src/agent/composer.ts

The guidance system is layered:

  • shared blocks
  • profile blocks
  • app-type blocks
  • stack blocks
  • option blocks
  • operational composition

Legacy AI Path

There is still a secondary AI path in:

  • src/ai/generator.ts
  • src/ai/provider.ts

This path can use an inference provider or a rule-based fallback. It is not the main generation flow anymore.