AGENTS.md

Visualize Me By Photo: AI Agent Context

Architecture Principles

• MVI with Circuit: Each feature is composed of Screen, Presenter, and UiState.
• DI: Dependency injection is managed via Metro.
• Stability: All major state classes must comply with compose-stability-analyzer for STABLE status.

Module Structure

• feature/*: Independent UI feature modules (Centered around Circuit Screen/Presenter).
• domain/*: Business logic, use cases, and domain entities (Pure Kotlin).
• data/*: Local (Room) and remote (Retrofit) data sources (Data mapping and repository).
• core/*: Shared components, design system, and utilities.

Global Rules for AI

1. Dependency Rule (Code Dependency Direction):

   • Communication must occur via domain interfaces or core modules.
   • All modules may depend on core.
   • feature and data modules may depend on domain.
   • domain must not depend on feature or data layers (Pure Kotlin logic).
   • Direct circular or peer dependencies between feature and data are strictly prohibited.

2. Data Flow (Data Transfer Path):

   • Data propagates from the data layer, through the domain layer, and ultimately to the feature layer.
   • Use Domain Models to decouple layers during data transfer; never leak DTOs outside the data layer.

3. Domain Isolation:

   • domain modules must only contain pure Kotlin logic, with no dependencies on Android or Compose frameworks.

4. Stability:

   • If compose-stability-analyzer reports stale data, use the --rerun-tasks option to force a baseline update.

5. Compose:

   • All UI must be written in a Declarative manner, and state transitions must be handled via CircuitEvent.

6. Code Generation:

   • Every generated file must end with a trailing newline.
   • No wildcard imports (import foo.* is banned).
   • Trailing comma required on both call site and declaration site.
   • Blank line required between package declaration and import block.