Skip to content

Strategic Plan: Kotlin Foundation Grant Winners Skills Integration in kmp-cli

Executive Summary

The Kotlin Foundation Grants Program identifies, funds, and accelerates the most critical open-source libraries across the Kotlin Multiplatform (KMP) ecosystem. Developers building production KMP applications frequently choose these grant-winning libraries for image rendering, caching, key-value storage, state architecture, rich text editing, Swift interop, web presentation, and cloud backends.

This plan details how kmp-cli will curate, package, and provide certified Agent Skills (SKILL.md adhering to the agentskills.io standard) for each official Kotlin Foundation Grant Winner library. By embedding these skills into kmp-cli, both AI coding assistants (Antigravity, Claude, Cursor) and human engineers will gain access to vetted multiplatform architecture recipes, dependency coordinates, and strict boundary hygiene directives.

1. Vision & Architectural Principles

1.1 Goals

  1. Canonical Authority: Every skill reflects the latest multiplatform idioms (Kotlin 2.1+, Compose Multiplatform 1.7+, Kotlin Toolchain module.yaml and Gradle KTS).
  2. Zero-Bloat & Boundary Hygiene: Strict separation of concerns — purely multiplatform abstractions in commonMain, with platform-specific engines encapsulated cleanly.
  3. Agentic Discoverability: Each skill contains trigger phrases, structured frontmatter, compatibility matrices, and anti-pattern checklists to guide autonomous coding agents.
  4. Seamless CLI Integration: Accessible via kmp skills search, kmp skills install, and kmp analyze libraries.

1.2 Hierarchy of Canonical Truth

In alignment with the KMP CLI Plugin governance: - Tier 1 (Contextual Truth): The target project's build files (build.gradle.kts / module.yaml, declared targets). - Tier 2 (Canonical Authority): Official Kotlin Foundation ecosystem guidelines, klibs.io metadata, and library author documentation. - Tier 3 (Procedural Runbooks): The certified SKILL.md recipes in skills/<skill-id>/SKILL.md.

2. Grant Winners Skills Inventory & Taxonomy

The 8 primary grant-winning libraries from 2023–2025 are organized into four functional domains:

flowchart TD
    subgraph KF["Kotlin Foundation Grant Winners"]
        Media["1. Media & Presentation"]
        Data["2. Data & Persistence"]
        Arch["3. Architecture & State"]
        Platforms["4. Platforms & Interop"]
    end

    Media --> Coil["Coil 3<br/><code>kotlin-media-coil</code>"]
    Media --> RichEditor["Compose Rich Editor<br/><code>kotlin-ui-compose-rich-editor</code>"]
    Media --> Kobweb["Kobweb<br/><code>kotlin-web-kobweb</code>"]

    Data --> Settings["Multiplatform-Settings<br/><code>kotlin-data-multiplatform-settings</code>"]
    Data --> Store["Store 5<br/><code>kotlin-data-store5</code>"]

    Arch --> Orbit["Orbit MVI<br/><code>kotlin-architecture-orbit-mvi</code>"]

    Platforms --> SwiftInterop["KMP-NativeCoroutines / SKIE<br/><code>kotlin-interop-swift-coroutines</code>"]
    Platforms --> Firebase["Firebase Kotlin SDK<br/><code>kotlin-cloud-firebase</code>"]

3. Detailed Skill Specifications

3.1 kotlin-media-coil (Coil 3)

  • Grant Year: 2023
  • Primary Maintainer: Colin White
  • Canonical Coordinates: io.coil-kt.coil3:coil-compose, io.coil-kt.coil3:coil-network-ktor3
  • Target Platforms: Android, iOS, Desktop (JVM), Web (WasmJs, JS)
  • Core Recipe:
  • Image loader configuration in commonMain using AsyncImage and SubcomposeAsyncImage.
  • Multiplatform network engine integration with Ktor 3 (KtorNetworkFetcherFactory).
  • Cross-platform disk caching with okio filesystem integration (FileSystem.SYSTEM_TEMPORARY_DIRECTORY).
  • Anti-Patterns:
  • Never use Android-only AndroidUri or context directly in commonMain composables.
  • Never block the UI thread with synchronous image decoding.

3.2 kotlin-data-multiplatform-settings (Multiplatform-Settings)

  • Grant Year: 2024
  • Primary Maintainer: Russell Wolf
  • Canonical Coordinates: com.russhwolf:multiplatform-settings, com.russhwolf:multiplatform-settings-coroutines
  • Target Platforms: Android, iOS, Desktop (JVM), Web (JS/WasmJs)
  • Core Recipe:
  • Factory pattern providing ObservableSettings across targets:
    • Android: SharedPreferencesSettings / DataStoreSettings.
    • iOS: NSUserDefaultsSettings.
    • Desktop: PreferencesSettings (Java Preferences API).
    • Web: StorageSettings (Browser LocalStorage).
  • Reactive Flow subscriptions using getIntFlow, getStringOrNullFlow.
  • Anti-Patterns:
  • Do not use for large relational datasets (direct agents to kotlin-data-room-kmp or kotlin-data-sqldelight-database instead).
  • Never store unencrypted auth tokens or cryptographic keys in standard settings (delegate to platform keychain/keystore).

3.3 kotlin-data-store5 (Store 5)

  • Grant Year: 2024
  • Primary Maintainer: Mobile Native Foundation (Mike Nakhimovich)
  • Canonical Coordinates: org.mobilenativefoundation.store:store5
  • Target Platforms: Android, iOS, Desktop, WasmJs
  • Core Recipe:
  • Offline-first single-source-of-truth orchestration.
  • Composing Fetcher.of { key -> ktorClient.fetch(key) } with SourceOfTruth.of(reader = { key -> sqldelight.select(key) }, writer = { key, data -> sqldelight.insert(key, data) }).
  • Consuming store.stream(StoreReadRequest.cached(key, refresh = true)) with typed StoreReadResponse.Data, Loading, and Error.
  • Anti-Patterns:
  • Bypassing the local cache for hot UI updates.
  • Mixing networking error handling directly inside the UI layer.

3.4 kotlin-architecture-orbit-mvi (Orbit MVI)

  • Grant Year: 2024
  • Primary Maintainer: Mikolaj Leszczynski / Babylon Health
  • Canonical Coordinates: org.orbit-mvi:orbit-core, org.orbit-mvi:orbit-compose
  • Target Platforms: Android, iOS, Desktop, WasmJs
  • Core Recipe:
  • ContainerHost<STATE, SIDE_EFFECT> pattern coupled with androidx.lifecycle.ViewModel.
  • Declarative state transformations: intent { reduce { state.copy(...) } }.
  • Non-persistent one-off event dispatch: postSideEffect(ToastEvent).
  • Deterministic testing harness using containerHost.test(initialState) { runIntent { ... } expectState { ... } }.
  • Anti-Patterns:
  • Firing side-effects directly inside reduce blocks (must remain pure).
  • Storing navigation state directly in the MVI state instead of delegating to the router.

3.5 kotlin-ui-compose-rich-editor (Compose Rich Editor)

  • Grant Year: 2024
  • Primary Maintainer: Mohamed Rejeb
  • Canonical Coordinates: com.mohamedrejeb.richeditor:richeditor-compose
  • Target Platforms: Android, iOS, Desktop, Web
  • Core Recipe:
  • Managing styled text through rememberRichTextState().
  • Providing toolbar controls (bold, italic, strikethrough, bullet points, headers, inline code).
  • Bi-directional export/import: HTML (state.toHtml(), state.setHtml(...)) and Markdown (state.toMarkdown()).
  • Anti-Patterns:
  • Reinventing custom span parsers with platform SpannableStringBuilder.
  • Mutating rich text state outside the Compose snapshot system.

3.6 kotlin-interop-swift-coroutines (KMP-NativeCoroutines & SKIE)

  • Grant Year: 2023 (KMP-NativeCoroutines by Rick Clephas) & 2024 (SKIE by Touchlab)
  • Canonical Coordinates: co.touchlab.skie:skie-gradle-plugin / com.rickclephas.kmp.nativecoroutines
  • Target Platforms: iOS, macOS, watchOS, tvOS
  • Core Recipe:
  • Seamless Swift modern concurrency bridging: mapping Kotlin StateFlow<T> directly to Swift AsyncSequence and SwiftUI @Observable.
  • Bridging Kotlin sealed interfaces to exhaustive Swift enums.
  • Proper cancellation propagation from Swift Task to Kotlin Job.
  • Anti-Patterns:
  • Forcing iOS engineers to write manual completion handlers or memory-leaking closures.
  • Ignoring Kotlin Coroutine cancellation when iOS view controllers dismiss.

3.7 kotlin-web-kobweb (Kobweb)

  • Grant Year: 2025
  • Primary Maintainer: David Herman (Varabyte)
  • Canonical Coordinates: com.varabyte.kobweb:kobweb-core, com.varabyte.kobweb:kobweb-silk
  • Target Platforms: Web (WasmJs, JS)
  • Core Recipe:
  • Full-stack Compose HTML development with file-based routing (@Page).
  • Responsive design tokens with Silk UI (Breakpoint, Modifier.fillMaxWidth()).
  • Server-side API endpoints (@Api) and static site export.
  • Anti-Patterns:
  • Mixing Android/Desktop Skiko Compose dependencies in Kobweb DOM modules.
  • Hardcoding absolute viewport dimensions instead of responsive Silk breakpoints.

3.8 kotlin-cloud-firebase (Firebase Kotlin SDK)

  • Grant Year: 2025
  • Primary Maintainer: GitLive
  • Canonical Coordinates: dev.gitlive:firebase-firestore, dev.gitlive:firebase-auth, dev.gitlive:firebase-storage
  • Target Platforms: Android, iOS, Desktop, Web
  • Core Recipe:
  • Idiomatic multiplatform Firebase client without platform-specific SDK leaks.
  • Coroutine-based document queries: Firebase.firestore.collection("users").document("id").snapshots.
  • Kotlinx Serialization integration for automatic document decoding: doc.data<UserProfile>().
  • Anti-Patterns:
  • Calling Google Play Services Task.await() in commonMain.
  • Writing raw Map/JSON parsing instead of @Serializable data classes.

4. Engineering Deliverables & Implementation Phases

The integration will be delivered across four structured phases:

Phase Scope Target Skills Deliverables
Phase 1: Persistence & Media High-frequency baseline needs for mobile and desktop apps kotlin-data-multiplatform-settings
kotlin-media-coil
kotlin-data-store5		 - 3 SKILL.md documents
- Entries in skills/catalog.json
- Entries in DefaultSkillsCatalog.kt
- Unit tests in SeedSkillsValidationTest
Phase 2: Architecture & Rich UI State management and text editing capabilities kotlin-architecture-orbit-mvi
kotlin-ui-compose-rich-editor		 - 2 SKILL.md documents
- Catalog synchronization
- Sample usage tests
Phase 3: Swift Interop & Cloud Native iOS ergonomics, web expansion, and BaaS kotlin-interop-swift-coroutines
kotlin-web-kobweb
kotlin-cloud-firebase		 - 3 SKILL.md documents
- Multiplatform interop recipes
- Catalog synchronization
Phase 4: CLI Enhancement Discovery and automatic curation CLI features - Add --tag grant-winner filter to kmp skills list
- Visual badge [GRANT WINNER] in kmp skills search
- Automatic recommendation in kmp analyze libraries

5. Catalog Schema & Metadata Specification

Each skill will be registered in skills/catalog.json and DefaultSkillsCatalog.kt using this schema:

{
  "id": "kotlin-media-coil",
  "name": "kotlin-media-coil",
  "description": "Multiplatform async image loading, caching, and Ktor 3 integration with Coil 3 for Compose Multiplatform.",
  "version": "1.0.0",
  "author": "jesusdmedinac",
  "category": "media",
  "aliases": [
    "kmp-coil",
    "coil3"
  ],
  "tags": [
    "coil",
    "images",
    "media",
    "cache",
    "grant-winner"
  ],
  "targets": [
    "android",
    "ios",
    "desktop",
    "wasm"
  ],
  "path": "skills/kotlin-media-coil/SKILL.md"
}

6. Verification & Quality Assurance

To ensure that each integrated skill meets the highest standards: 1. Frontmatter & Markdown Linting: Ensure each SKILL.md parses cleanly via SkillParser.kt. 2. Catalog Integrity Test: Extend SeedSkillsValidationTest.kt to assert the presence of all 8 grant-winner IDs, aliases, and tags (grant-winner). 3. Target Compatibility: Cross-reference dependency coordinates and compatibility matrices against klibs.io API. 4. Zero-Bloat Verification: Ensure sample snippets compile under both standard Gradle KTS and declarative module.yaml.

7. Open Questions & Prioritization

[!NOTE] Review the following strategic decisions before kickoff:

  1. Phase 1 Prioritization: Would you prefer starting immediately with Phase 1 (multiplatform-settings, coil3, store5), or should we prioritize Swift interop (SKIE / KMP-NativeCoroutines) first?
  2. Naming Convention: Do you approve the kotlin-<category>-<library> standard (e.g. kotlin-media-coil, kotlin-data-multiplatform-settings), with short aliases (e.g. coil, settings) for fast CLI invocation?
  3. CLI Badging: Should kmp skills list and kmp analyze libraries explicitly flag libraries as [Kotlin Foundation Grant Winner] in terminal outputs?