Android logo

Android

Community
ayush016
android

Senior-level Android engineering expertise for Jetpack Compose, Material 3, shared element transitions, navigation, performance, security, background work, notifications, adaptive layouts, Coil image loading, WorkManager, FCM, Baseline Profiles, R8 optimization, accessibility, coroutines and Flow, and extraordinary UI quality. Load this skill for ANY Android or Kotlin task — building a screen, animating a transition, designing a data layer, scaffolding a multi-module project, optimising startup, adding biometric auth, writing WorkManager jobs, wiring FCM, handling foldables, writing Roborazzi screenshot tests, debugging on device with ADB or MCP. If it touches Android, this skill covers it. Non-negotiable aesthetic bar: every screen produced must look like it belongs in a design award showcase. Default grey scaffolds, missing animations, hardcoded colors, and spinners as sole loading states are bugs, not style choices. Trigger on: Kotlin, Jetpack Compose, ViewModel, Hilt, Room, Retrofit, Coroutines, Flow, Navigation, WorkManager, FCM, Coil, Roborazzi, Macrobenchmark, Baseline Profiles, R8, WindowSizeClass, foldables, shared element transitions, animation, ADB, MCP.

Overview

Publisherayush016
Repositoryandroid-lead-agent-skills
Skill nameandroid
Stars
65
Forks
8
Bundled files
20
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.

  • 20 bundled files

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

  • Open source

    Published by ayush016 on GitHub. Read the source before you install it.

Installation

Install the Android 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/ayush016/android-lead-agent-skills/tree/main
  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/ayush016/android-lead-agent-skills.git \
  .claude/skills/android
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Android 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 Android 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 Android 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.

Android Development Skill

Production-grade Android engineering expertise from data layer to pixel, from startup milliseconds to biometric auth, from phone portrait to foldable landscape. UI quality, performance, security, and accessibility are first-class concerns equal to functional correctness.


⚡ Quick Decision Guide

TaskReference
Color theming, 29 roles, token hierarchy, gradientstheming-and-color.md
Building a new screencompose-ui-system.md
Navigation with shared element transitionsshared-element-transitions.md
Type-safe navigation, deep links, multi-modulenavigation.md
Animation / motion (non-shared-element)motion-and-animation.md
Adaptive layouts, large screens, foldablesadaptive-layouts.md
Coroutines, Flow, dispatchers, cancellationcoroutines-and-flow.md
Room, Retrofit, offline-first data layerdata-layer.md
ViewModel, state, DI, architecturearchitecture.md
Baseline Profiles, R8, recomposition perfperformance.md
Encryption, biometrics, cert pinningsecurity.md
WorkManager, Foreground Services, Alarmsbackground-work.md
Coil3, SubcomposeAsyncImage, cachingimage-loading.md
FCM, notification channels, deep linksnotifications.md
TalkBack, semantics, focus, contrastaccessibility.md
Unit, screenshot, Roborazzi teststesting.md
Convention plugins, version catalog, modulesbuild-and-modules.md
ADB / Android MCP debuggingandroid-mcp.md
Pre-ship quality reviewassets/ui-excellence-checklist.md

Core Engineering Principles

1. Unidirectional Data Flow, No Exceptions State flows down from ViewModel. Events flow up via lambdas or actions. No composable reads from a database. No ViewModel imports Compose. The boundary is sharp and testable.

2. Clean Architecture as a Dependency Rule Domain has zero Android imports. Data is invisible to UI. Features depend on core, never each other. Violations create debt that compounds faster than feature velocity.

3. UI Excellence Is Not Optional Producing a visually mediocre screen is an engineering failure. Default grey scaffolds, hardcoded colors, missing transitions, and spinners as primary loading states are incomplete work.

4. Shared Element Transitions Are the Default for Content Navigation Tapping a card must cause the card to flow into its detail. sharedBounds() or sharedElement() is the default. Opt out only when there is genuinely no spatial relationship to express.

5. Physics Over Timing spring() over tween() for gesture-coupled and state-change animations. Tune stiffness and dampingRatio deliberately — defaults are starting points, not final choices.

6. The 4dp Spacing Grid Is a Contract Every spacing value is a multiple of 4dp, referenced via AppSpacing.* tokens. Arbitrary values are bugs that accumulate into visual dissonance.

7. Fakes Over Mocks Test ViewModels against in-memory fake implementations. Fakes exercise real contracts, survive refactors. Mocks couple tests to implementation details and silently pass when production code breaks.

8. Performance Is a Feature Baseline Profiles are committed for every app shipped. Recomposition counts are checked with Layout Inspector before any UI work is done. Cold start is tracked in CI via Macrobenchmark.

9. Security from Day One Encrypted storage for all sensitive data. Certificate pinning for all network endpoints. No secrets in source code. Input validated at every external boundary. These are not features to add later.

10. Accessibility Is Not Optional Every image has a content description. Every custom component declares its semantic role. Every interactive element has a 48dp touch target. Colour contrast meets WCAG AA. TalkBack is tested before any screen is shipped.


★ THE EXTRAORDINARY UI MANDATE ★

Android users have seen a million apps. The ones they remember have something alive about them — a transition that makes content feel like a physical object moving through space, surfaces that catch light at different elevations, a loading state so well crafted it does not feel like waiting.

The Bar. Every screen passes three tests before it ships:

  1. The scroll-stop test. Would a designer pause on this if scrolling past it? Not because it is garish — because something is considered. A hierarchy that breathes. A transition that reveals spatial relationship. A surface treatment that says "we care."

  2. The feel test. Does it respond to touch like a physical object? Spring physics on press, haptic coordination with state changes, skeleton loaders that mirror real content — these are what separate an app people recommend from one people tolerate.

  3. The motion test. Do transitions communicate where content came from? Shared elements communicate spatial memory. Instant cuts communicate nothing. Slide transitions say you moved sideways. Shared elements say: this is that thing, grown.

In code, this means:

  • Shared element transitions for every navigation between related content — always
  • spring() physics tuned for each context; no unexamined defaults
  • Color exclusively from Material 3 roles; Brush gradients where flat color is visually weak
  • A display typeface for headlines paired with a humanist sans for body
  • Skeleton screens mirroring the exact geometry of real content — no centred spinners
  • Edge-to-edge always; status bar and nav bar are part of the design
  • Canvas for anything standard components cannot achieve
  • Haptic feedback coordinated with meaningful state transitions

Anti-patterns to refuse: Default grey scaffold backgrounds. Hardcoded hex colors in composables. Instant navigation between related content. LinearProgressIndicator as the sole loading state. Cards with identical elevation everywhere. Spacing not on the 4dp grid.


Shared Element Transitions — Overview

SharedTransitionLayout coordinates geometry between composable trees during navigation. sharedElement() matches identical content (image in list → same image in detail). sharedBounds() matches composables sharing a spatial region (card → full screen).

Three things must thread down: SharedTransitionScope, AnimatedVisibilityScope, and a stable entity-ID key. Missing any one silently breaks the transition.

Full guide, four complete Kotlin patterns, pitfalls: references/shared-element-transitions.md


Android MCP Integration

MCP = eyes on device. Code = hands in codebase. Build → install → screenshot → compare → refine. Catches what Roborazzi misses: real inset behaviour, animation timing, system bar styling, density quirks.

Full capability table, workflows, log filtering, hierarchy inspection: references/android-mcp.md


Feature Build Workflow

  1. Module — Create with yourapp.android.library.compose plugin. build-and-modules.md
  2. Routes — @Serializable data class in :core:navigation. navigation.md
  3. UI State — Sealed interface: Loading, Ready(data), Error(message, canRetry). architecture.md
  4. Actions — Sealed interface for all user interactions.
  5. ViewModel — StateFlow + SharedFlow(replay=0) + SavedStateHandle.toRoute(). architecture.md
  6. Domain — UseCase → Repository interface → impl in :core:data. data-layer.md
  7. Composables — Public screen + private sub-composables; AppSpacing, colorScheme, typography only. compose-ui-system.md
  8. Shared elements — sharedElement() or sharedBounds() with stable ID keys. shared-element-transitions.md
  9. Motion — Stagger entry, spring state changes, enter/exit transitions. motion-and-animation.md
  10. Skeleton loading — Mirror Ready geometry with ShimmerBox. compose-ui-system.md
  11. Adaptive — Verify on 600dp+; apply WindowSizeClass-aware layout. adaptive-layouts.md
  12. Accessibility — Semantics, content descriptions, focus order, 48dp targets. accessibility.md
  13. Security — No secrets; input validated at all external boundaries. security.md
  14. MCP review — Screenshot in light/dark/large font/RTL; check insets + transitions. android-mcp.md
  15. Tests — ViewModel unit tests + Roborazzi screenshots. testing.md
  16. UI excellence — Every item in assets/ui-excellence-checklist.md must pass.

Common Pitfalls

1. Hardcoded colors — Color.White in a composable. Fix: all colors via MaterialTheme.colorScheme.*.

2. Missing modifier parameter — Composable cannot be resized by caller. Fix: every public composable takes modifier: Modifier = Modifier on its outermost layout.

3. SharedFlow vs StateFlow for events — Toast fires again on rotation. Fix: events = MutableSharedFlow(replay=0). State = MutableStateFlow(initial).

4. Shared element key mismatch — Navigation works but no transition fires. Fix: build from stable entity ID. Log both sides and compare.

5. SharedTransitionLayout not wrapping NavHost — IllegalStateException: No SharedTransitionScope. Fix: SharedTransitionLayout { NavHost { } } — must contain both source and destination.

6. Missing renderInSharedTransitionScope — Back button blinks into existence at transition end. Fix: apply .renderInSharedTransitionScope(scope).animateEnterExit(...) to destination-only elements.

7. Blocking main thread — ANR or UI freeze. Fix: withContext(dispatchers.io) for all I/O; never runBlocking on the main dispatcher.

8. Spinner as only loading state — CircularProgressIndicator centred on a blank background. Fix: ShimmerBox composable mirroring Ready geometry.

9. Non-4dp spacing — Visual rhythm feels slightly off; design review flags it. Fix: AppSpacing.* for every spacing value.

10. collectAsState instead of collectAsStateWithLifecycle — Flow collects in background. Fix: always collectAsStateWithLifecycle() for UI-bound collection.

11. GlobalScope usage — Coroutine leaks, ignores cancellation. Fix: viewModelScope, lifecycleScope, or rememberCoroutineScope(). Never GlobalScope.

12. Images loaded at full resolution for thumbnails — OOM in lists. Fix: .size(width, height) in ImageRequest matching display dimensions.

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 Android AI skill do?

Senior-level Android engineering expertise for Jetpack Compose, Material 3, shared element transitions, navigation, performance, security, background work, notifications, adaptive layouts, Coil image loading, WorkManager, FCM, Baseline Profiles, R8 optimization, accessibility, coroutines and Flow, and extraordinary UI quality. Load this skill for ANY Android or Kotlin task — building a screen, animating a transition, designing a data layer, scaffolding a multi-module project, optimising startup, adding biometric auth, writing WorkManager jobs, wiring FCM, handling foldables, writing Roboraz...

Why use Android on TypingMind?

Because you install it once and use it with any model. Android 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 Android in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/ayush016/android-lead-agent-skills/tree/main. 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 Android?

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 Android?

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

Is the Android AI skill free?

Yes. It is published on GitHub by ayush016 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 👇