← Browse

@haidrrrry/compose-kotlin-agent-skills

A

AGENTS.md — compose-kotlin-agent-skills

instructionscodex

Install

agr install @haidrrrry/compose-kotlin-agent-skills --target codex

Writes 1 file into AGENTS.md, pinned to git-f4672e9d.

  • AGENTS.md

Document

AGENTS.md — compose-kotlin-agent-skills

Enterprise Android/Kotlin agent skill kit. Plain-text index for agentic IDEs (Cursor, Claude Code, Codex, Gemini, Copilot, Windsurf, OpenHands, and 20+ others).

Repository: https://github.com/haidrrrry/compose-kotlin-agent-skills
Primary skill: SKILL.md — read this first for every Kotlin/Android task.

Platforms: Android · Kotlin Multiplatform · Compose Multiplatform (target) · macOS / Linux / Windows (host) · 27 agents.

Build & Validation

# Validate all SKILL.md files (frontmatter + links) — must pass before merge
python3 scripts/validate_skills.py

# Strict mode + lock parity + JUnit report (matches CI exactly)
python3 scripts/validate_skills.py \
  --strict \
  --lock-check api/skills.lock \
  --junit junit-results.xml

# When you add or remove a SKILL.md, regenerate the lock file:
python3 scripts/validate_skills.py --write-lock api/skills.lock

CI runs the same script on every push and pull request (.github/workflows/skill-lint.yml) and uploads junit-results.xml as an artifact.

Skill Catalog

SkillPathLoad When
compose-kotlin-agent-skills (primary)SKILL.mdAny Android/Kotlin/Compose work — toolchain, MVI guardrails, banned antipatterns
android-kotlin-architectureskills/android-kotlin-architecture/SKILL.mdClean Architecture, MVVM/MVI, modules, UseCases, UiState/Event/Effect
android-kotlin-composeskills/android-kotlin-compose/SKILL.mdCompose UI, recomposition, Material 3, edge-to-edge, animations, performance
android-kotlin-testingskills/android-kotlin-testing/SKILL.mdViewModel tests, Compose UI tests, Hilt fakes, Turbine, CI quality gates

Reference Modules (deep dives)

Load from references/ after reading the matching skill above. One level deep only — do not chain reference-to-reference.

Architecture & Platform

ReferenceTopic
references/00-banned-antipatterns.mdMaster banned table — WRONG/RIGHT lookup
references/01-architecture.mdClean Arch, MVVM, MVI, module structure
references/04-coroutines-flow.mdStateFlow, structured concurrency, errors
references/05-hilt-di.mdHilt, scopes, EntryPoint, testing
references/06-room-db.mdRoom, migrations, offline-first
references/07-navigation.mdNavigation 3, type-safe routes, deep links
references/08-kmp-cmp.mdKMP, expect/actual, shared ViewModel
references/09-networking.mdKtor, JWT, DTO mappers
references/14-datastore.mdDataStore, SharedPreferences migration
references/15-paging3.mdPaging 3, RemoteMediator, asState()
references/18-gradle-build-logic.mdConvention plugins, KSP, R8, baseline profiles
references/19-xml-to-compose-migration.mdXML → Compose, RxJava → Flow

Compose & Performance

ReferenceTopic
references/02-compose-ui.mdComposition, stability, Modifier order, theming
references/03-animations.mdSprings, Canvas, gestures, shared elements
references/10-performance.mdRecomposition, LazyColumn, baseline profiles
references/16-coil-image.mdCoil 3.4, AsyncImage, cache
references/17-accessibility.mdTalkBack, semantics, WCAG

Quality & Ship

ReferenceTopic
references/11-testing.mdTurbine, Compose tests, Hilt TestInstallIn
references/12-camera-mlkit.mdCameraX, ML Kit pose, angle math
references/13-release-checklist.mdSigning, R8, Play Store

Agent Install Guides (27 platforms)

Directory: agents/ · Index: agents/README.md · Paste snippet: agents/_shared-snippet.md

git clone https://github.com/haidrrrry/compose-kotlin-agent-skills.git <agent-dir>/skills/compose-kotlin-agent-skills

Replace <agent-dir> per agent (.cursor, .claude, .codex, .windsurf, etc.) — see the matching file in agents/.

Production Code Examples

ExampleSource
examples/todo-mvi/MVI todo list — ViewModel, Screen, Repository (antipattern-safe reference)
examples/animated-clock/AnimatedClockJetpacl — Canvas, CMP, particles
examples/authenticator/Authenticator — Room, MVVM, Koin, theming

What This Repository Is

Agent skill kit = markdown instructions for coding AIs (Cursor, Claude Code, Codex, etc.) — not an Android library, not Alexa/Bixby, not an on-device LLM SDK.

Install: git clone into .cursor/skills/compose-kotlin-agent-skills or .claude/skills/compose-kotlin-agent-skills.

Not the same as Compose UI libraries (coachmark / onboarding / theming Composables you add via Gradle). Those ship UI in your app. This repo teaches your coding agent how to write Kotlin/Compose correctly.

Boundaries — Do Not

  • Confuse this repo with a Gradle dependency, voice-assistant skill, or Compose UI library
  • Invent dependency versions — use the version table in SKILL.md
  • Use GlobalScope, naked _state.value =, or hardcoded UI strings
  • Skip LazyColumn keys or use collectAsState() on Android screens
  • Merge without passing python3 scripts/validate_skills.py --strict --lock-check api/skills.lock
  • Add reference files without linking them from SKILL.md or this index
  • Add a new SKILL.md without bumping api/skills.lock and the catalog table above

Verification Prompt

Read AGENTS.md and SKILL.md. Structure a Navigation 3 screen with MVI UiState,
edge-to-edge insets, and collectAsStateWithLifecycle. Cite banned antipatterns you avoided.

Expected: Kotlin 2.x, AGP 9, _state.update { }, stringResource, NavKey routes, no GlobalScope.

Trustgrade A

  • passBody integrity

    Whether the stored document is plausibly the kind of file the artifact declares, rather than something fetched by mistake.

  • passType matchnot applicable to this artifact type

    Whether the artifact is really the kind of thing its metadata claims it is.

  • passFreshness

    How long since the source repository was last pushed to.

  • passPrompt injection

    Scans the artifact's own text for instructions aimed at your agent rather than at you.

  • passLicense

    Whether the source repository declares an SPDX license permissive enough to redistribute.

How the grade is calculated

Each check contributes 0 points when it passes, 1 when it warns, and 2 when it fails. The total maps to a letter:

  • Aevery check passed
  • Bone warning
  • Ctwo warnings
  • Dprompt injection or body integrity failed, or three warnings
  • Fone of those failed, and something else is wrong

These are automated hygiene checks, not a security audit, and not a dependency or vulnerability scan. A grade of A means nothing was flagged — not that the artifact is safe.

Versions

  • git-f4672e9d33ea2026-08-04