Kmp Compose Multiplatform logo

Kmp Compose Multiplatform

Community
felipechaux
kmp-compose-multiplatform

Expert Kotlin Multiplatform (KMP) and Compose Multiplatform development guidance. Use when creating, reviewing, or modifying KMP projects with Compose UI, clean architecture, and multi-platform targets (Android, iOS, Desktop, Web).

Overview

Publisherfelipechaux
Repositorykmp-compose-multiplatform-skill
Skill namekmp-compose-multiplatform
Stars
83
Forks
6
Bundled files
8
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 8 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by felipechaux on GitHub. Read the source before you install it.

Installation

Install the Kmp Compose Multiplatform AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/felipechaux/kmp-compose-multiplatform-skill/tree/main/.claude/skills/kmp-compose-multiplatform
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/felipechaux/kmp-compose-multiplatform-skill.git /tmp/kmp-compose-multiplatform-skill
mkdir -p .claude/skills
cp -r /tmp/kmp-compose-multiplatform-skill/.claude/skills/kmp-compose-multiplatform .claude/skills/kmp-compose-multiplatform
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Kmp Compose Multiplatform in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Kmp Compose Multiplatform on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Kmp Compose Multiplatform is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Kotlin Multiplatform + Compose Multiplatform Skill

You are an expert in Kotlin Multiplatform (KMP) and Compose Multiplatform development. You follow Google's official architecture guidelines (as demonstrated in Now in Android), JetBrains Compose best practices, and the KMP community standards.

Core Principles

  1. Maximize shared code — write once in commonMain, use everywhere
  2. Clean Architecture — strict layer separation: Data → Domain → Presentation
  3. Feature-based modularization — organize by feature, not by layer
  4. Unidirectional Data Flow (UDF) — state flows down, events flow up
  5. Interface-first design — define contracts, inject implementations
  6. Platform parity — same behavior on Android and iOS unless explicitly platform-specific

Project Structure

Recommended Module Layout

root/
├── app/                          # Android app entry point
├── iosApp/                       # iOS app entry point (Xcode project)
├── shared/                       # KMP shared module (or multi-module)
│   └── src/
│       ├── commonMain/           # Shared code for all platforms
│       ├── androidMain/          # Android-specific implementations
│       ├── iosMain/              # iOS-specific implementations
│       └── commonTest/           # Shared tests
├── build-logic/                  # Convention plugins (if multi-module)
│   └── convention/               # Gradle convention plugins
└── gradle/
    └── libs.versions.toml        # Version catalog (ALWAYS use this)

Feature Module Layout (inside commonMain)

Each feature must follow this exact structure:

feature/
└── [feature-name]/
    ├── data/
    │   ├── local/
    │   │   ├── dao/              # Room DAOs
    │   │   └── entity/           # Room entities
    │   ├── remote/               # API services
    │   ├── repository/           # Repository implementations
    │   └── mapper/               # Data ↔ Domain mappers
    ├── domain/
    │   ├── model/                # Domain models (pure Kotlin)
    │   ├── repository/           # Repository interfaces
    │   └── usecase/              # Use cases (one action per class)
    ├── presentation/
    │   ├── ui/                   # Composable screens and components
    │   ├── viewmodel/            # ViewModels
    │   └── state/                # UI state data classes
    └── di/                       # Koin module for this feature

Architecture Guidelines

Layer Responsibilities

Data Layer

  • Implements repository interfaces from domain
  • Maps data models to/from domain models
  • Handles network requests (Ktor) and local persistence (Room/DataStore)
  • Never exposes data models to domain or presentation

Domain Layer

  • Pure Kotlin — NO Android/platform dependencies
  • Repository interfaces (abstractions)
  • Use cases: single public function operator fun invoke()
  • Domain models (not database entities, not DTOs)

Presentation Layer

  • ViewModels hold StateFlow<UiState> — never expose mutable state
  • UI State is a sealed class or data class
  • Composables receive state + callbacks (no direct ViewModel access in nested composables)
  • Navigation handled at screen level only

Resource/Result Pattern

Always use a sealed class for async results with typed domain errors (never raw strings):

kotlin
sealed class Resource<out T> {
    data class Success<out T>(val data: T) : Resource<T>()
    data class Error(val error: AppError) : Resource<Nothing>()
    data object Loading : Resource<Nothing>()
}

See references/error-handling.md for the full AppError hierarchy, safeApiCall wrapper, and error-to-UI-message mapping.

Use Case Pattern

kotlin
class GetUserUseCase(private val repository: UserRepository) {
    suspend operator fun invoke(userId: String): Resource<User> {
        return repository.getUser(userId)
    }
}

ViewModel Pattern

Use data class UiState (not sealed class) for composable state with _uiState.update { }. Expose navigation events via a separate SharedFlow:

kotlin
class HomeViewModel(
    private val getItemsUseCase: GetItemsUseCase
) : ViewModel() {

    private val _uiState = MutableStateFlow(HomeUiState())
    val uiState: StateFlow<HomeUiState> = _uiState.asStateFlow()

    // One-time navigation/event channel — never put navigation in UiState
    private val _events = MutableSharedFlow<HomeEvent>()
    val events: SharedFlow<HomeEvent> = _events.asSharedFlow()

    fun loadItems() {
        viewModelScope.launch {
            _uiState.update { it.copy(isLoading = true, errorMessage = null) }
            getItemsUseCase()
                .onSuccess { items ->
                    _uiState.update { it.copy(isLoading = false, items = items) }
                }
                .onError { error ->
                    _uiState.update { it.copy(isLoading = false, errorMessage = error.toUserMessage()) }
                }
        }
    }

    fun onItemClicked(id: String) {
        viewModelScope.launch {
            _events.emit(HomeEvent.NavigateToDetail(id))
        }
    }
}

// Flat data class — preferred over sealed class for composable state
data class HomeUiState(
    val isLoading: Boolean = false,
    val items: List<Item> = emptyList(),
    val errorMessage: String? = null   // human-readable, never AppError
)

// One-time events — navigation, toasts, analytics
sealed class HomeEvent {
    data class NavigateToDetail(val id: String) : HomeEvent()
    data object ShowUndoSnackbar : HomeEvent()
}

Collect events in the screen composable:

kotlin
@Composable
fun HomeScreen(
    navController: NavHostController,
    viewModel: HomeViewModel = koinViewModel()
) {
    val uiState by viewModel.uiState.collectAsStateWithLifecycle()

    // Collect one-time events
    LaunchedEffect(Unit) {
        viewModel.events.collect { event ->
            when (event) {
                is HomeEvent.NavigateToDetail -> navController.navigate(Screen.Detail.createRoute(event.id))
                is HomeEvent.ShowUndoSnackbar -> { /* show snackbar */ }
            }
        }
    }

    HomeContent(uiState = uiState, onItemClick = viewModel::onItemClicked)
}

StateFlow from Repository Flow

Use stateIn() to convert a repository Flow into a ViewModel StateFlow:

kotlin
val uiState: StateFlow<HomeUiState> = itemsRepository.observeItems()
    .map { items -> HomeUiState(items = items) }
    .stateIn(
        scope = viewModelScope,
        started = SharingStarted.WhileSubscribed(5_000),
        initialValue = HomeUiState(isLoading = true)
    )

Kotlin Multiplatform Patterns

Expect/Actual Pattern

Use expect/actual for platform-specific implementations:

kotlin
// commonMain
expect fun getPlatformName(): String

expect class DatabaseBuilder(context: Any?) {
    fun build(): AppDatabase
}
kotlin
// androidMain
actual fun getPlatformName(): String = "Android"

actual class DatabaseBuilder actual constructor(private val context: Any?) {
    actual fun build(): AppDatabase =
        Room.databaseBuilder(context as Context, AppDatabase::class.java, "app.db").build()
}
kotlin
// iosMain
actual fun getPlatformName(): String = "iOS"

actual class DatabaseBuilder actual constructor(context: Any?) {
    actual fun build(): AppDatabase {
        val dbFilePath = NSHomeDirectory() + "/app.db"
        return Room.databaseBuilder<AppDatabase>(name = dbFilePath).build()
    }
}

Source Set Configuration (build.gradle.kts)

kotlin
kotlin {
    androidTarget {
        compilations.all {
            compileTaskProvider.configure {
                compilerOptions {
                    jvmTarget.set(JvmTarget.JVM_17)
                }
            }
        }
    }

    listOf(iosX64(), iosArm64(), iosSimulatorArm64()).forEach { target ->
        target.binaries.framework {
            baseName = "shared"
            isStatic = true
        }
    }

    sourceSets {
        commonMain.dependencies {
            // Compose Multiplatform
            implementation(compose.runtime)
            implementation(compose.foundation)
            implementation(compose.material3)
            implementation(compose.ui)
            implementation(compose.components.resources)

            // Navigation
            implementation(libs.navigation.compose)

            // Koin
            implementation(libs.koin.core)
            implementation(libs.koin.compose)
            implementation(libs.koin.compose.viewmodel)

            // Ktor
            implementation(libs.ktor.client.core)
            implementation(libs.ktor.client.content.negotiation)
            implementation(libs.ktor.serialization.kotlinx.json)

            // Room
            implementation(libs.room.runtime)
            implementation(libs.room.ktx)

            // DataStore
            implementation(libs.datastore.preferences)

            // DateTime
            implementation(libs.kotlinx.datetime)

            // Serialization
            implementation(libs.kotlinx.serialization.json)

            // Coroutines
            implementation(libs.kotlinx.coroutines.core)
        }

        androidMain.dependencies {
            implementation(libs.ktor.client.okhttp)
            implementation(libs.koin.android)
            implementation(libs.kotlinx.coroutines.android)
        }

        iosMain.dependencies {
            implementation(libs.ktor.client.darwin)
        }
    }
}

Dependency Injection with Koin

Module Structure

kotlin
// feature/home/di/HomeModule.kt
val homeModule = module {
    single<HomeRepository> { HomeRepositoryImpl(get(), get()) }
    factory { GetHomeDataUseCase(get()) }
    viewModel { HomeViewModel(get()) }
}

Scopes — Feature-Scoped Dependencies

Use Koin scopes for dependencies that should live only as long as a feature/screen is active (e.g., a shopping cart, a multi-step form):

kotlin
// Define a scope qualifier
val CartScope = named("CartScope")

val cartModule = module {
    // Scoped — one instance per CartScope lifecycle
    scope(CartScope) {
        scoped { CartRepository(get()) }
        scoped { CartViewModel(get()) }
    }
}

// Open scope when entering the feature
val cartScope = getKoin().createScope("cart_session", CartScope)
val cartViewModel = cartScope.get<CartViewModel>()

// Close scope when leaving — instance is garbage collected
cartScope.close()

Lazy Injection

Use inject() (lazy delegation) instead of get() (eager) when the dependency may not be needed immediately:

kotlin
class HomeViewModel : ViewModel() {
    private val analyticsService: AnalyticsService by inject()   // lazy
    private val repository: HomeRepository = get()               // eager
}

Named Qualifiers

Use named() qualifiers when you need multiple instances of the same type in the same module — a common pattern for multiple API clients or dispatchers:

kotlin
val networkModule = module {
    // Two HTTP clients with different base URLs, distinguished by name
    single<HttpClient>(named("main")) {
        provideHttpClient(baseUrl = BuildKonfig.API_BASE_URL, tokenProvider = get())
    }
    single<HttpClient>(named("auth")) {
        provideHttpClient(baseUrl = BuildKonfig.AUTH_BASE_URL, tokenProvider = get())
    }

    // Multiple dispatchers
    single<CoroutineDispatcher>(named("io")) { Dispatchers.IO }
    single<CoroutineDispatcher>(named("main")) { Dispatchers.Main }
}

// Inject by name
class UserRepository(
    private val mainClient: HttpClient = get(named("main")),
    private val authClient: HttpClient = get(named("auth"))
)

ViewModel with SavedStateHandle

Bind SavedStateHandle in Koin using viewModelOf or the params API:

kotlin
// Using viewModelOf — automatically injects SavedStateHandle
val featureModule = module {
    viewModelOf(::DetailViewModel)  // SavedStateHandle injected automatically
}

// Or manually via params
val featureModule = module {
    viewModel { params ->
        DetailViewModel(
            savedStateHandle = params.get(),
            getItemUseCase = get()
        )
    }
}
kotlin
class DetailViewModel(
    savedStateHandle: SavedStateHandle,
    private val getItemUseCase: GetItemUseCase
) : ViewModel() {
    private val itemId: String = checkNotNull(savedStateHandle[Screen.Detail.ARG_ID])
}

Central Module Aggregator

kotlin
// di/AppModule.kt
fun getAllModules() = listOf(
    platformModule(),
    coreModule,
    authModule,
    homeModule,
    // ... other feature modules
)

// Platform-specific (expect/actual)
expect fun platformModule(): Module

Koin Initialization

kotlin
// KoinInitializer.kt
fun initKoin(appDeclaration: KoinAppDeclaration = {}) {
    startKoin {
        appDeclaration()
        modules(getAllModules())
    }
}

Android entry (Application class):

kotlin
class MyApp : Application() {
    override fun onCreate() {
        super.onCreate()
        initKoin {
            androidContext(this@MyApp)
        }
    }
}

iOS entry (Swift):

swift
KoinInitializerKt.doInitKoin()

Build System

Version Catalog (gradle/libs.versions.toml)

Always use the version catalog. Never hardcode versions in build files:

toml
[versions]
kotlin = "2.3.0"
compose-multiplatform = "1.10.1"
agp = "8.8.0"
koin = "4.1.1"
ktor = "3.0.3"
room = "2.8.4"
datastore = "1.1.1"
navigation-compose = "2.9.1"
kotlinx-coroutines = "1.10.2"
kotlinx-serialization = "1.7.3"
kotlinx-datetime = "0.6.1"
ksp = "2.3.0-1.0.32"
buildkonfig = "0.17.1"
coil = "3.0.4"

[libraries]
# Koin
koin-core = { module = "io.insert-koin:koin-core", version.ref = "koin" }
koin-android = { module = "io.insert-koin:koin-android", version.ref = "koin" }
koin-compose = { module = "io.insert-koin:koin-compose", version.ref = "koin" }
koin-compose-viewmodel = { module = "io.insert-koin:koin-compose-viewmodel", version.ref = "koin" }
# Ktor
ktor-client-core = { module = "io.ktor:ktor-client-core", version.ref = "ktor" }
ktor-client-okhttp = { module = "io.ktor:ktor-client-okhttp", version.ref = "ktor" }
ktor-client-darwin = { module = "io.ktor:ktor-client-darwin", version.ref = "ktor" }
ktor-client-content-negotiation = { module = "io.ktor:ktor-client-content-negotiation", version.ref = "ktor" }
ktor-serialization-kotlinx-json = { module = "io.ktor:ktor-serialization-kotlinx-json", version.ref = "ktor" }
# Room
room-runtime = { module = "androidx.room:room-runtime", version.ref = "room" }
room-ktx = { module = "androidx.room:room-ktx", version.ref = "room" }
room-compiler = { module = "androidx.room:room-compiler", version.ref = "room" }
# DataStore
datastore-preferences = { module = "androidx.datastore:datastore-preferences-core", version.ref = "datastore" }
# Navigation
navigation-compose = { module = "org.jetbrains.androidx.navigation:navigation-compose", version.ref = "navigation-compose" }
# KotlinX
kotlinx-coroutines-core = { module = "org.jetbrains.kotlinx:kotlinx-coroutines-core", version.ref = "kotlinx-coroutines" }
kotlinx-coroutines-android = { module = "org.jetbrains.kotlinx:kotlinx-coroutines-android", version.ref = "kotlinx-coroutines" }
kotlinx-serialization-json = { module = "org.jetbrains.kotlinx:kotlinx-serialization-json", version.ref = "kotlinx-serialization" }
kotlinx-datetime = { module = "org.jetbrains.kotlinx:kotlinx-datetime", version.ref = "kotlinx-datetime" }
# Coil
coil-compose = { module = "io.coil-kt.coil3:coil-compose", version.ref = "coil" }
coil-network-ktor = { module = "io.coil-kt.coil3:coil-network-ktor3", version.ref = "coil" }

[plugins]
kotlin-multiplatform = { id = "org.jetbrains.kotlin.multiplatform", version.ref = "kotlin" }
compose-multiplatform = { id = "org.jetbrains.compose", version.ref = "compose-multiplatform" }
compose-compiler = { id = "org.jetbrains.kotlin.plugin.compose", version.ref = "kotlin" }
android-library = { id = "com.android.library", version.ref = "agp" }
android-application = { id = "com.android.application", version.ref = "agp" }
kotlin-serialization = { id = "org.jetbrains.kotlin.plugin.serialization", version.ref = "kotlin" }
ksp = { id = "com.google.devtools.ksp", version.ref = "ksp" }
room = { id = "androidx.room", version.ref = "room" }
buildkonfig = { id = "com.codingfeline.buildkonfig", version.ref = "buildkonfig" }

BuildKonfig for Environment Configuration

kotlin
// build.gradle.kts
buildkonfig {
    packageName = "com.example.shared"

    defaultConfigs {
        buildConfigField(STRING, "ENVIRONMENT", "stage")
        buildConfigField(BOOLEAN, "IS_DEBUG", "true")
        buildConfigField(STRING, "API_BASE_URL", "https://api.stage.example.com")
    }

    targetConfigs("prod") {
        buildConfigField(STRING, "ENVIRONMENT", "prod")
        buildConfigField(BOOLEAN, "IS_DEBUG", "false")
        buildConfigField(STRING, "API_BASE_URL", "https://api.example.com")
    }
}

Data Persistence

Room Database Setup

kotlin
// commonMain
@Database(entities = [UserEntity::class], version = 2)
abstract class AppDatabase : RoomDatabase() {
    abstract fun userDao(): UserDao
}
kotlin
// androidMain — actual
actual class DatabaseBuilder actual constructor(private val context: Any?) {
    actual fun build(): AppDatabase = Room.databaseBuilder<AppDatabase>(
        context = context as Context,
        name = context.getDatabasePath("app.db").absolutePath
    )
    .addMigrations(MIGRATION_1_2)
    .build()
}

val MIGRATION_1_2 = object : Migration(1, 2) {
    override fun migrate(db: SupportSQLiteDatabase) {
        db.execSQL("ALTER TABLE users ADD COLUMN avatar_url TEXT")
    }
}
kotlin
// iosMain — actual
actual class DatabaseBuilder actual constructor(context: Any?) {
    actual fun build(): AppDatabase = Room.databaseBuilder<AppDatabase>(
        name = NSHomeDirectory() + "/app.db"
    )
    .addMigrations(MIGRATION_1_2)
    .build()
}

Room DAO — Reactive Queries

Always use Flow<List<T>> for queries that the UI observes — never return a raw List:

kotlin
@Dao
interface UserDao {
    // Reactive — emits whenever the table changes
    @Query("SELECT * FROM users ORDER BY name ASC")
    fun observeAll(): Flow<List<UserEntity>>

    // One-shot suspend for writes
    @Upsert
    suspend fun upsert(user: UserEntity)

    @Delete
    suspend fun delete(user: UserEntity)

    @Query("SELECT * FROM users WHERE id = :id")
    suspend fun getById(id: String): UserEntity?

    // Transaction for atomic multi-step operations
    @Transaction
    suspend fun replaceAll(users: List<UserEntity>) {
        deleteAll()
        insertAll(users)
    }

    @Query("DELETE FROM users")
    suspend fun deleteAll()

    @Insert(onConflict = OnConflictStrategy.IGNORE)
    suspend fun insertAll(users: List<UserEntity>)
}

Room Pagination with Paging 3

For large datasets use PagingSource — never load everything into memory:

toml
# libs.versions.toml
paging = "3.3.6"
[libraries]
paging-runtime = { module = "androidx.paging:paging-runtime", version.ref = "paging" }
paging-compose = { module = "androidx.paging:paging-compose", version.ref = "paging" }
paging-testing = { module = "androidx.paging:paging-testing", version.ref = "paging" }
kotlin
// DAO — return PagingSource instead of List
@Dao
interface ItemDao {
    @Query("SELECT * FROM items ORDER BY created_at DESC")
    fun pagingSource(): PagingSource<Int, ItemEntity>
}

// Repository
fun observeItemsPaged(): Flow<PagingData<Item>> = Pager(
    config = PagingConfig(pageSize = 20, enablePlaceholders = false),
    pagingSourceFactory = { itemDao.pagingSource() }
).flow.map { pagingData -> pagingData.map { it.toDomain() } }

// ViewModel
val pagedItems: Flow<PagingData<Item>> = itemsRepository
    .observeItemsPaged()
    .cachedIn(viewModelScope)

// Composable
@Composable
fun ItemListScreen(viewModel: HomeViewModel = koinViewModel()) {
    val items = viewModel.pagedItems.collectAsLazyPagingItems()

    LazyColumn {
        items(count = items.itemCount, key = items.itemKey { it.id }) { index ->
            items[index]?.let { ItemCard(item = it) }
        }
        item {
            when (items.loadState.append) {
                is LoadState.Loading -> CircularProgressIndicator()
                is LoadState.Error -> RetryButton(onClick = { items.retry() })
                else -> Unit
            }
        }
    }
}

Room Full-Text Search (FTS)

kotlin
@Fts4(contentEntity = ItemEntity::class)
@Entity(tableName = "items_fts")
data class ItemFtsEntity(
    @PrimaryKey @ColumnInfo(name = "rowid") val rowId: Int = 0,
    val title: String,
    val description: String
)

@Dao
interface ItemSearchDao {
    @Query("SELECT * FROM items WHERE rowid IN (SELECT rowid FROM items_fts WHERE items_fts MATCH :query)")
    fun search(query: String): Flow<List<ItemEntity>>
}

DataStore Setup

kotlin
// commonMain
expect fun createDataStore(producePath: () -> String): DataStore<Preferences>

internal const val DATASTORE_FILE = "app_prefs.preferences_pb"

Networking with Ktor

kotlin
// commonMain
class ApiService(private val client: HttpClient) {

    suspend fun getUser(id: String): UserDto =
        client.get("/users/$id").body()

    suspend fun createUser(request: CreateUserRequest): UserDto =
        client.post("/users") {
            contentType(ContentType.Application.Json)
            setBody(request)
        }.body()
}

// HTTP client setup (in DI module) — with auth, logging, and retry
fun provideHttpClient(
    baseUrl: String,
    tokenProvider: TokenProvider
): HttpClient = HttpClient {
    install(ContentNegotiation) {
        json(Json {
            ignoreUnknownKeys = true
            isLenient = true
        })
    }
    install(HttpTimeout) {
        requestTimeoutMillis = 30_000
        connectTimeoutMillis = 15_000
        socketTimeoutMillis = 30_000
    }
    install(Logging) {
        logger = Logger.DEFAULT
        level = if (BuildKonfig.IS_DEBUG) LogLevel.HEADERS else LogLevel.NONE
    }
    // Automatic retry for transient failures
    install(HttpRequestRetry) {
        retryOnServerErrors(maxRetries = 3)
        retryOnException(maxRetries = 3, retryOnTimeout = true)
        exponentialDelay(base = 2.0, maxDelayMs = 10_000)
    }
    // Auth token injection
    install(Auth) {
        bearer {
            loadTokens { BearerTokens(tokenProvider.getAccessToken(), tokenProvider.getRefreshToken()) }
            refreshTokens {
                val newTokens = tokenProvider.refresh()
                BearerTokens(newTokens.accessToken, newTokens.refreshToken)
            }
        }
    }
    defaultRequest {
        url(baseUrl)
        header(HttpHeaders.ContentType, ContentType.Application.Json)
    }
}

Always wrap API calls with safeApiCall to map Ktor exceptions to domain errors — see references/error-handling.md.

Exponential Backoff with Jitter

Add randomization to retry delays to prevent thundering-herd problems when many clients retry simultaneously:

kotlin
install(HttpRequestRetry) {
    retryOnServerErrors(maxRetries = 3)
    retryOnException(maxRetries = 3, retryOnTimeout = true)
    // Add jitter: randomize delay within ±500ms of calculated backoff
    exponentialDelay(base = 2.0, maxDelayMs = 10_000, randomizationMs = 500)
}

Certificate Pinning (Android)

For high-security apps, pin the server's certificate to prevent MITM attacks:

kotlin
// androidMain — OkHttp CertificatePinner
actual fun createHttpClient(baseUrl: String, tokenProvider: TokenProvider): HttpClient =
    HttpClient(OkHttp) {
        engine {
            config {
                certificatePinner(
                    CertificatePinner.Builder()
                        .add("api.example.com", "sha256/AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=")
                        .add("api.example.com", "sha256/BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB=") // backup pin
                        .build()
                )
            }
        }
        // ... other config
    }

Keep two pins active at all times (primary + backup) to allow certificate rotation without a forced update.

SharedFlow Buffer Strategy

Choose buffer size and overflow behavior explicitly when emitting from multiple coroutines:

kotlin
// One-time UI events (navigation, toasts) — no replay, drop oldest if consumer is slow
private val _events = MutableSharedFlow<HomeEvent>(
    replay = 0,
    extraBufferCapacity = 64,
    onBufferOverflow = BufferOverflow.DROP_OLDEST
)

// App-wide events (logout, session expiry) — replay 1 so late subscribers catch the event
private val _appEvents = MutableSharedFlow<AppEvent>(
    replay = 1,
    extraBufferCapacity = 16,
    onBufferOverflow = BufferOverflow.DROP_OLDEST
)

Never use replay > 0 for navigation events — a screen re-subscribing would navigate again.

HTTP Caching

Enable response caching in the Ktor client to reduce network calls and support offline reading:

kotlin
// androidMain — OkHttp cache
actual fun createHttpClient(baseUrl: String, tokenProvider: TokenProvider): HttpClient =
    HttpClient(OkHttp) {
        engine {
            config {
                cache(Cache(
                    directory = context.cacheDir.resolve("http_cache"),
                    maxSize = 10L * 1024 * 1024  // 10 MB
                ))
            }
        }
        // ... other plugins
    }

For stale-while-revalidate behaviour, add headers in repository calls:

kotlin
suspend fun getItems(): Resource<List<ItemDto>> = safeApiCall {
    client.get("/items") {
        header(HttpHeaders.CacheControl, "max-age=300")   // fresh for 5 min
    }.body()
}

OAuth 2.0 Token Refresh

The Ktor Auth plugin handles token rotation automatically. Ensure the refresh call itself is unauthenticated to avoid infinite loops:

kotlin
install(Auth) {
    bearer {
        loadTokens {
            BearerTokens(tokenStorage.accessToken, tokenStorage.refreshToken)
        }
        refreshTokens {
            // markAsRefreshTokenRequest() prevents Auth plugin re-intercepting this call
            val response = client.post("/auth/refresh") {
                markAsRefreshTokenRequest()
                setBody(RefreshRequest(oldTokens?.refreshToken ?: ""))
            }.body<TokenResponse>()

            tokenStorage.save(response.accessToken, response.refreshToken)
            BearerTokens(response.accessToken, response.refreshToken)
        }
        sendWithoutRequest { request ->
            request.url.host == "api.example.com"   // only attach token to your API
        }
    }
}

---

## Internationalization (i18n)

All user-facing strings must use Compose Multiplatform's resource system. Never hardcode text:

```kotlin
// GOOD — uses generated Res.string references
Text(text = stringResource(Res.string.home_title, userName))
Button(onClick = onRetry) { Text(text = stringResource(Res.string.action_retry)) }

// BAD — hardcoded, not translatable
Text(text = "Welcome, $userName")

Key rules:

  • Define all strings in commonMain/composeResources/values/strings.xml
  • Add locale folders (values-es/, values-ar/) for each supported language
  • Use pluralStringResource() for quantities — never if (count == 1) string branching
  • Use start/end padding (not left/right) for RTL language support
  • Use Icons.AutoMirrored.* for directional icons that should flip in RTL
  • Test with @Preview(locale = "ar") to verify RTL layouts

See references/i18n.md for plurals, RTL testing, dynamic locale change, and locale-aware number/currency formatting.


Testing Strategy

Unit Tests (commonTest)

kotlin
class GetUserUseCaseTest {

    private val repository = FakeUserRepository()
    private val useCase = GetUserUseCase(repository)

    @Test
    fun `returns success when repository succeeds`() = runTest {
        repository.setUser(testUser)
        val result = useCase("user-123")
        assertIs<Resource.Success<User>>(result)
        assertEquals(testUser, result.data)
    }
}

// Fake (not mock) — real implementation of the interface
class FakeUserRepository : UserRepository {
    private var user: User? = null

    fun setUser(user: User) { this.user = user }

    override suspend fun getUser(id: String): Resource<User> =
        user?.let { Resource.Success(it) } ?: Resource.Error("Not found")
}

Rules:

  • Never use Mockito or MockK — use fakes/test doubles
  • All shared tests go in commonTest
  • Platform-specific tests in androidTest/iosTest
  • Use runTest from kotlinx-coroutines-test for coroutine testing

Logging

Use expect/actual for platform logging — never use println() in production code. Never log sensitive data (tokens, passwords, PII).

kotlin
// commonMain — log levels matching platform conventions
enum class LogLevel { DEBUG, INFO, WARN, ERROR }

expect fun logDebug(tag: String, message: String)
expect fun logInfo(tag: String, message: String)
expect fun logWarn(tag: String, message: String)
expect fun logError(tag: String, message: String, throwable: Throwable? = null)
kotlin
// androidMain — use Timber for structured Android logging
actual fun logDebug(tag: String, message: String) = Timber.tag(tag).d(message)
actual fun logInfo(tag: String, message: String) = Timber.tag(tag).i(message)
actual fun logWarn(tag: String, message: String) = Timber.tag(tag).w(message)
actual fun logError(tag: String, message: String, throwable: Throwable?) =
    Timber.tag(tag).e(throwable, message)

Initialize Timber in Application.onCreate() — plant a DebugTree for debug builds and a Crashlytics reporting tree for production:

kotlin
class MyApp : Application() {
    override fun onCreate() {
        super.onCreate()
        if (BuildKonfig.IS_DEBUG) {
            Timber.plant(Timber.DebugTree())
        } else {
            Timber.plant(CrashlyticsTree())
        }
    }
}

// Production tree — routes WARN/ERROR to Firebase Crashlytics
class CrashlyticsTree : Timber.Tree() {
    override fun log(priority: Int, tag: String?, message: String, t: Throwable?) {
        // Only forward warnings and errors to crash reporting
        if (priority < Log.WARN) return
        // Never log sensitive data — scrub before sending
        val safeMessage = message.redactSensitivePatterns()
        FirebaseCrashlytics.getInstance().log("[$tag] $safeMessage")
        if (t != null) FirebaseCrashlytics.getInstance().recordException(t)
    }
}

// Redact tokens, emails, phone numbers before sending to crash services
private fun String.redactSensitivePatterns(): String = this
    .replace(Regex("Bearer [A-Za-z0-9\\-._~+/]+=*"), "Bearer [REDACTED]")
    .replace(Regex("[a-zA-Z0-9._%+\\-]+@[a-zA-Z0-9.\\-]+\\.[a-zA-Z]{2,}"), "[EMAIL REDACTED]")
kotlin
// iosMain — use os_log (not NSLog, which is deprecated for structured logging)
import platform.Foundation.NSLog
import platform.darwin.*

actual fun logDebug(tag: String, message: String) {
    os_log_with_type(OS_LOG_DEFAULT, OS_LOG_TYPE_DEBUG, "[$tag] $message")
}
actual fun logInfo(tag: String, message: String) {
    os_log_with_type(OS_LOG_DEFAULT, OS_LOG_TYPE_INFO, "[$tag] $message")
}
actual fun logWarn(tag: String, message: String) {
    os_log_with_type(OS_LOG_DEFAULT, OS_LOG_TYPE_ERROR, "[$tag] WARN: $message")
}
actual fun logError(tag: String, message: String, throwable: Throwable?) {
    os_log_with_type(OS_LOG_DEFAULT, OS_LOG_TYPE_FAULT, "[$tag] ERROR: $message ${throwable?.message ?: ""}")
}

Common Pitfalls to Avoid

  1. Never put Android/iOS imports in commonMain — use expect/actual
  2. Never expose Flow from Room directly to UI — map through repository to domain models first
  3. Never use LiveData in KMP — use StateFlow/Flow only
  4. Never hardcode strings in Compose — use stringResource() from compose resources
  5. Never use rememberCoroutineScope in a ViewModel — use viewModelScope
  6. Never pass Context through layers — inject at the platform module level only
  7. Never use GlobalScope — use structured concurrency with viewModelScope or CoroutineScope(SupervisorJob())
  8. Avoid LaunchedEffect for ViewModel operations — use collectAsStateWithLifecycle()
  9. Never share mutable state across composables — hoist to a single source of truth
  10. Do not skip the domain layer — even for simple features, maintain the abstraction
  11. Never use raw String for errors in Resource.Error — use typed AppError sealed class
  12. Never put navigation calls in UiState — use a separate SharedFlow<Event> for one-time events
  13. Never call stopKoin() in production code — only in test teardown
  14. Never use println() for logging — use expect/actual log functions
  15. Never skip Room migrations — always add a Migration object when bumping the schema version
  16. Never omit contentDescription on meaningful images/icons — required for accessibility (TalkBack, VoiceOver)
  17. Never log sensitive data — redact tokens, emails, and PII before sending to Crashlytics or any log aggregation service
  18. Never load unbounded lists — use PagingSource + Pager for large datasets
  19. Never hardcode user-facing strings — always use stringResource() from compose resources
  20. Never use left/right padding in Composables — use start/end for RTL language support
  21. Never use Icons.Default.ArrowBack for navigation — use Icons.AutoMirrored.Filled.ArrowBack to mirror in RTL
  22. Never use reply > 0 on navigation event SharedFlows — late subscribers would trigger navigation again
  23. Never keep only one certificate pin — always pin primary + backup to allow rotation without forcing an update
  24. Never use left/right in column/row alignment — prefer Start/End which respect layout direction

Reference Files

  • references/architecture.md — detailed architecture guide, module structures, feature flags, inter-feature communication, proto DataStore
  • references/compose-best-practices.md — composable design, @Stable, state hoisting, Material 3, focus management, text field accessibility, dynamic type, previews, performance
  • references/error-handling.md — AppError hierarchy, safeApiCall, recoverable vs fatal, 429 handling, error analytics/breadcrumbs, retry logic
  • references/testing.md — fakes, ViewModel tests with Turbine, SharedFlow event testing, Paging tests, screenshot/golden tests, Compose UI tests, Room in-memory
  • references/ios-interop.md — Swift naming conventions, SKIE sealed class edge cases, Kotlin/Native memory model, iOS performance, nullability bridging, coroutines↔Swift Concurrency
  • references/navigation.md — deep links, cross-module navigation contracts, predictive back, deep link validation, nested nav, bottom navigation, back handling, transitions, SavedStateHandle
  • references/build-system.md — convention plugins, R8/ProGuard, publishing to Maven, CI Gradle daemon, KSP config, gradle.properties, build performance
  • references/i18n.md — string resources, plurals, RTL support, dynamic locale change, locale-aware number/currency formatting

Official References

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Kmp Compose Multiplatform AI skill do?

Expert Kotlin Multiplatform (KMP) and Compose Multiplatform development guidance. Use when creating, reviewing, or modifying KMP projects with Compose UI, clean architecture, and multi-platform targets (Android, iOS, Desktop, Web).

Why use Kmp Compose Multiplatform on TypingMind?

Because you install it once and use it with any model. Kmp Compose Multiplatform is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Kmp Compose Multiplatform in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/felipechaux/kmp-compose-multiplatform-skill/tree/main/.claude/skills/kmp-compose-multiplatform. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Kmp Compose Multiplatform?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Kmp Compose Multiplatform?

As many as you like. As long as a model supports skills, you can use Kmp Compose Multiplatform with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Kmp Compose Multiplatform AI skill free?

Yes. It is published on GitHub by felipechaux under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇