Skip to content

Agent.md

Latest commit

 

History

History
49 lines (37 loc) · 2.91 KB

Agent.md

File metadata and controls

49 lines (37 loc) · 2.91 KB
Raw

AGENTS Guidelines for This Repository

This repository contains an Android application project. When working on the project interactively with an AI coding agent, please follow the guidelines below to ensure architectural consistency, maximum performance, and a smooth development experience.

1. Project Specifications

  • Minimum SDK: 24 (or defined by project)
  • Target SDK: 34
  • Language: Kotlin (1.9+)
  • Build System: Gradle (Kotlin DSL preferred)

2. Architecture & Design Patterns

We follow the official Android Architecture Guidelines:

  • Presentation Layer: StateFlow/SharedFlow in ViewModels. UI defined in Jetpack Compose (or XML if legacy). Unidirectional Data Flow (UDF).
  • Domain Layer: Optional UseCases for complex business logic.
  • Data Layer: Repository pattern to abstract data sources (Room for local, Retrofit for remote).
  • Dependency Injection: Hilt/Dagger (preferred) or Koin.

3. Asynchronous Programming

  • Concurrency: Kotlin Coroutines exclusively. Avoid RxJava for new code (migrate if possible).
  • Dispatchers: Inject Dispatchers (don't hardcode Dispatchers.IO) to allow testing.

4. UI Framework

  • Jetpack Compose: Default choice for all new features. Follow Compose best practices (state hoisting, modifiers, no side-effects in composables).
  • Navigation: Jetpack Navigation Compose.

5. Testing Philosophy

  • Unit Tests: JUnit4/JUnit5, MockK for mocking, Turbine for Flow testing.
  • UI Tests: Compose Test Rule for UI components, Espresso for legacy XML.
  • Prefer testing ViewModel state emission over testing implementation details.

6. External Documentation

7. Useful Agent Skills Recap

Skill Folder Purpose
architecture/ Clean architecture, ViewModels, and Data Layer.
ui/ Jetpack Compose best practices, Coil, Accessibility.
performance/ Auditing Compose and Gradle build performance.
migration/ XML to Compose, RxJava to Coroutines.
testing_and_automation/ Unit/UI Testing setup, Emulator automation scripts.
concurrency_and_networking/ Coroutines fixes, Retrofit networking.

Following these practices ensures that the agent-assisted development workflow stays reliable and consistent. When in doubt, always refer to the specific agent skills provided in .github/skills/ for deeper task-specific context!

Note to developers: Update this file whenever the project makes architectural shifts to ensure AI agents stay aligned with your conventions.