Let AI agents generate a production grade Flutter codebase
AI agents impose their own defaults instead of following proven architectures. Flutter Agent Kit gives them battle tested rules and feature skills so they ship code that fits.
Context layer
Your agent gets the architecture, not just feature recipes
Even without a specific feature skill, the agent has the architecture, data patterns, service conventions and UI rules to build any feature solidly. On-demand docs give it exactly the context it needs, pulled in only when the work calls for them so the prompt stays lean.
- architecture.mdApp initialization and state management
- data.mdModel conventions and storage layer
- services.mdHTTP services and service patterns
- ui.mdNavigation, theming and shared widgets
- conventions.mdNaming conventions and imports
- setup.mdProject setup and FVM configuration
- git.mdGit workflow and branching strategy
- process.mdChange tracking and development process
Works with
Skills
Feature workflows your agent can add without being an expert in each
Each skill encodes a production hardened pattern for a common feature, so the agent ships code that follows the kit's architecture instead of improvising its own. See every feature.
- gorouter.mdRouting configuration with named routes and deep links
- local-db.mdSembast store with CRUD services and a v3 Notifier backed by the DB
- firestore.mdCloud Firestore collections with real-time streams
- firebase-auth.mdLogin, signup, session state and social sign-in through Firebase Auth
- revenuecat.mdPaywalls, subscriptions and per-feature gating through RevenueCat
- admob.mdBanner, interstitial and rewarded ads with AdMob integration
- in-app-review.mdIn-app review prompt with rating flow and throttling
- push-notifications.mdPush opt-in, token sync and deep-link routing on tap
- ai-chat.mdStreaming and tool calling against OpenAI-compatible endpoints
- i18n.mdTranslations and locale switching with opt-in slang codegen
- sentry.mdCrash reporting with breadcrumbs and user context
- firebase-analytics.mdEvent tracking and screen views with Firebase Analytics
- dialogs.mdReactive prompts, confirmations and error toasts
- onboarding.mdFirst-launch intro that gates the home screen until complete
- settings.mdUser options, about section, legal links and a debug env switcher
- google-maps.mdGoogle Maps widget with markers, polylines and camera control
- camera.mdCamera capture and image picker from gallery with permissions handled
- web.mdWeb target support with a capability layer and hosting templates
- ui-design.mdNative-fidelity design bar with an anti-slop pre-flight for screens
About
Matteo Ricupero
Software developer for over 10 years. Flutter expert, shipped apps for Scalapay, Akiflow, InsideOutJapan and more plus a dozen personal production apps. CS graduate, University of Insubria.
More at ricu.it →In production
Battle tested across shipping apps, not toy demos.
Some of personal production apps. Many more have been built for clients, distilled into the kit.
| Any AI for Notion | AI assistant for Notion workspaces. |
| BitPong | Unofficial FITET client for Italian table tennis data. |
| Pola Browser | Privacy-focused mobile browser. |
| TT Club Manager | Table tennis club management app. |
| RememberFast Spaced Repetition | SuperMemo 2 flashcard app for language learning. |
Shipped
BitPong, shipped with the patterns in this kit
Unofficial FITET client for Italian table tennis data. 1,000+ active users on the App Store.
Problem
Every new Flutter project starts the same.
- 01Explain the architecture.
- 02Explain the folder structure.
- 03Explain your coding conventions.
- 04Explain Riverpod.
- 05Explain routing.
- 06Explain dependency injection.
Flutter Agent Kit gives your AI coding agent a proven architecture to follow
Built from 10+ years of software development experience
Instead of inventing patterns from scratch, your coding agent gets rules, structure and architectural decisions already made. Clear conventions, battle tested patterns and a coherent foundation to generate code that fits your project without creating chaos.
AI coding agents are fast.
Without a solid architecture, they're fast at creating technical debt.
Pricing
One payment, every future skill. No recurring fees, no per-seat licensing.
- Lifetime updates
- One-time, no fees
- Unlimited projects
What's included
- Scaffold command + feature-first Riverpod v3 template
- AGENTS.md + docs (architecture, data, services, UI, conventions)
- 19 production skills (auth, payments, push, AI chat, maps, more)
- Lifetime updates included with your purchase
- Works with Cursor, Claude Code, opencode, Copilot, Gemini
- Unlimited personal and commercial projects
FAQ
Questions before you buy