A comprehensive library of modular AI Skills for Flutter & Dart development, designed for precise, production-ready AI Agent workflows. Each skill provides structured knowledge and expert-level guidance optimized for LLM consumption and human readability.
All skills follow a modular, flat structure for maximum clarity and AI compatibility.
skills/
βββ <skill-name>/
βββ SKILL.md # Entry point: Goals, Process, Constraints
βββ references/ # Detailed implementation guides
- Single Responsibility: Each skill focuses on one technology or framework
- LLM-Optimized: Documents are structured for AI context windows with semantic headings
- Production-Ready: Working code snippets, not placeholders
- Version-Aware: All packages use latest stable versions (as of 2026-03-31)
flutter-provider: Provider (v6.1.5+1) - ChangeNotifier + MVVM with memory leak prevention and disposal patternsflutter-riverpod: Riverpod (v3.3.1) - Riverpod 3.0 compile-safe reactive state;Consumerscoped rebuilds; generic providers; mutations & offline persistence (experimental)flutter-bloc: BLoC (v9.1.1) - Event-driven architecture with transformers and Freezed integration
flutter-shared-preferences: SharedPreferences (v2.5.5) - Simple key-value for non-sensitive settingsflutter-secure-storage: SecureStorage (v10.0.0) - Encrypted storage with biometric support, iOS Keychain & Android Keystoreflutter-hive: Hive CE (v2.19.x) - High-performance NoSQL object storageflutter-drift: Drift (v2.32.x) - Type-safe reactive SQL with WAL mode and sqlite3 v3.x
flutter-gorouter: GoRouter (v17.1.0) - Official declarative routing with deep linking and redirect loop preventionflutter-autoroute: AutoRoute (v11.1.0) - Type-safe routing with code generation
flutter-social-auth: OAuth/SSO for Google, Apple, Facebook, LINE with error handling and common gotchasflutter-deeplink: App Links and Universal Links configuration
flutter-expert: Architecture decision matrix, Clean/Feature-First patterns, performance optimizationflutter-responsive: Multi-screen layouts with breakpointsflutter-animate: High-performance animations and physics-based effectsflutter-hooks: Functional widget lifecycle managementflutter-isolate: Concurrency with ReceivePort cleanup and memory leak preventionflutter-testing: Unit, widget, integration, golden tests with CI/CD troubleshootingflutter-genui: AI-powered UI generationmarionette: AI-driven Flutter E2E testing via VM Service CLI β tap, scroll, enterText, screenshot on live debug builds
shadcn-flutter: Modern accessible UI components with themingeffective-dart: Dart 3.10 best practices β dot shorthands, null-aware collection elements, wildcard variables, records, patterns, sealed classes
fl-chart: FL Chart (v1.2.0) β Line, Bar, Pie, Scatter, Radar, Candlestick charts with implicit animations, touch tooltips, and pan/zoom transformations
supabase: Realtime Postgres, Edge Functions, Authenticationserverpod: Full-stack Dart server with type-safe endpointsfirebase: Analytics, Crashlytics, Authentication, Storage
freezed: Immutable models and union types with pattern matchingfpdart: Functional programming (Option, Either, Task)openapi-to-dart: Manual Dart API client implementation from OpenAPI specs using Dio + Equatable/Freezedts-to-dart: TypeScript to Dart conversion utilitiessentry-flutter: Error tracking and performance monitoringrevenuecat-flutter: Subscription billing and payment integrationflutter-ads: AdMob monetization and mediation
jaspr: Dart-only SSR/SPA web framework
flutter-setup: Environment setup with PATH configuration and troubleshooting for macOS/Linux/Windowsgithub-actions: CI/CD automation workflowsfastlane: Automated deployment pipelinescodemagic: Cloud CI/CD for Flutter apps
"Simplicity is the ultimate sophistication."
- Minimalist Approach: Focus on user experience, not engineering abstractions
- Modern Best Practices: Latest stable releases with version documentation
- Type Safety: Leverage Dart 3+ features (pattern matching, records, sealed classes)
- Zero Placeholder Policy: All code snippets are production-ready
- β 6 chart types: LineChart, BarChart, PieChart, ScatterChart, RadarChart, CandlestickChart
- β Quick start examples for all chart types with production-ready code
- β Common config patterns: titles/axes, grid, border, extra lines, touch interactivity, animations, pan/zoom
- β Advanced patterns: stacked bars with per-layer tooltip, multi-series gradient lines, donut chart touch expansion, multi-dataset radar overlay, horizontal bar rotation
- β 7 reference files: full API property tables for each chart type + shared base components
- β 36 SKILL.md files audited and standardized to Agent Skill Best Practices
- β
Gerund naming enforced (
applying-*,managing-*,testing-*) - β Third-person descriptions with trigger keywords for precise AI activation
- β
500-line limit enforced β
flutter-bloc(594β147 lines) andflutter-riverpod(625β162 lines) split with newreferences/files
- β
Breaking changes:
==equality for all state comparisons, fresh Notifier instances on rebuild (resource leak risk),FamilyNotifierremoved,Refgenerics removed - β
Consumer vs ConsumerWidget: decision table; prefer
Consumer+ref.watch().select()for surgical rebuilds - β
New APIs: generic providers,
@Dependenciesscoping,AsyncValuesealed +progress, weak listeners,overrideWithBuild - β
Mutations (experimental):
tsx.get()pattern, exhaustive 4-state switch (Idle/Pending/Error/Success) - β
Offline Persistence (experimental):
riverpod_sqflite,persist()inbuild(),StorageOptions
- β
Dot shorthands (3.10):
.blueinstead ofColor.blue,.all(16)instead ofEdgeInsets.all(16) - β
Null-aware collection elements (3.8):
[?nullable]auto-excludes nulls - β
Wildcard variables (3.7): non-binding
_, multiple_in same scope - β
New lints (3.9):
switch_on_type,unnecessary_unawaited,@awaitNotRequired
- β
Removed code-generation framing; now covers Dio client setup, Equatable vs Freezed decision table, Service/Repository layers,
DioException β ApiErrormapping
Each skill is self-contained and can be used independently. AI agents can load specific skills based on task requirements for precise, expert-level guidance.
---
name: "skill-name"
description: "Detailed description with trigger keywords for AI agents"
metadata:
last_modified: "YYYY-MM-DD HH:MM:SS (GMT+8)"
---
## Goal
Clear objective and use cases
## Process
Step-by-step implementation guide
## Reference Documentation
Links to detailed guides in references/
## Constraints
Critical rules and best practicesAll skills follow standardized formats optimized for LLM consumption. When adding new skills:
- Use single responsibility principle
- Include version-specific package information
- Provide working code examples
- Add trigger keywords in description for precise AI activation
Important
All documents follow the metadata convention:
metadata.last_modified: "YYYY-MM-DD HH:MM:SS (GMT+8)"