Ship a real app fast with AI
An Expo boilerplate built for Cursor and Claude Code. Open it and your AI sees the whole app: navigation, state, feature folders, and the rules it has to follow. So it writes clean, shippable React Native from the first prompt instead of guessing and breaking things.
Expo SDK 55, the New Architecture, RevenueCat, Firebase, Supabase, and the U-AMOS 2.0 memory system that keeps the AI consistent. Wired with strict TypeScript.
Built by Malik Chohra (9 years of mobile engineering), written for the way AI tools reason about code.
What's in the box
Every line below is checked against the source code, not marketing copy.
- Expo SDK 55 + React Native 0.83.6 with the New Architecture enabled
- TypeScript in strict mode with @/ path aliases pre-configured
- Authentication: Apple, Google, Email/Password, Forgot Password, biometric (Face ID, Touch ID, Fingerprint)
- RevenueCat paywall + free trial + restore purchases, A/B-testable across plans
- Firebase: Analytics, Crashlytics, Remote Config, Cloud Messaging
- Sentry bundled for richer error context alongside Crashlytics
- Offline-first todos (reference feature): MMKV cache, optimistic UI, sync queue, conflict resolution
- Internationalization: English, French, Spanish (auto-detected, easily extensible)
- Gamification: streaks, badges, points, leaderboard, Skia celebration animations
- Onboarding: configurable questionnaires, A/B variants, pre-paywall FOMO screen
- EAS Build with development / staging / production profiles + EAS Submit
- Jest + Testing Library for unit and integration tests
- U-AMOS 2.0: 9-file memory bank + 10+ rule packs + 97+ AI articles + step-by-step generators
Looking for cloud + on-device AI providers, multimodal analyser, and AI-generated UI? See the AI Pro tier →
Why your AI writes better code here
A blank Expo project gives Cursor nothing to anchor to. It guesses your folder structure, invents navigation patterns, and stitches together imports that compile but break at runtime. A smarter prompt won't fix that. A structure the AI can't misread will.
This codebase ships .cursorrules, CLAUDE.md, and a 9-file memory system tiered Hot, Warm, and Cold so your assistant loads only what it needs. Every feature folder has its own README. The UI library has an inventory the AI checks before referencing a component. The services have a registry. And 10+ rule packs the AI has to load before touching navigation, state, offline sync, or i18n.
The result: your AI references real components with real props, follows the same Redux Toolkit patterns the rest of the app uses, respects the offline-first sync model, and stops inventing files that don't exist. Vibe-coding speed, with code you can actually ship.
Get access
Beta is open. Drop your name, email, and GitHub username, and repository access lands in your inbox within 24 to 48 hours.
Request Beta Access
All three fields are required. We'll add you to the private GitHub repo.
Expo boilerplate FAQ
What problem does this Expo boilerplate actually solve?
AI tools write React Native code that looks right but breaks on mobile. They miss navigation lifecycles, leak memory across tabs, ignore platform differences, and use inconsistent state patterns. This codebase tells Cursor, Claude Code, and Antigravity exactly how your app is built, so the code fits: strict TypeScript, feature-first folders, explicit rules, and a memory system the AI reads before writing.
Which versions does it target?
Expo SDK 55.0.17 with the New Architecture enabled, React Native 0.83.6, TypeScript strict mode. Updated alongside each major Expo SDK release.
How many languages are included out of the box?
Three: English, French, and Spanish (en.json, fr.json, es.json under src/locales/). The device locale is auto-detected by expo-localization on first launch. Adding more languages is a one-file change.
Is end-to-end testing included?
No. The Standard tier ships Jest plus Testing Library for unit and integration tests, with coverage already in place for auth, todos, and notifications. End-to-end frameworks like Detox are not bundled. Add them if you need them.
Are AI provider integrations (OpenAI, Gemini, llama.rn) included here?
Not in this tier. The Standard boilerplate ships with zero AI provider integrations on purpose. It's a clean foundation. AI providers (Gemini and OpenAI in the cloud, llama.rn on-device, multimodal analyser, AI-generated UI) ship in the AI Pro tier.
Can I use this for a SaaS mobile app?
Yes. RevenueCat handles subscriptions and trials, Firebase handles analytics and feature flags, Supabase handles auth and Postgres with row-level security, and the feature-first architecture scales as your team grows.
Does it include Supabase?
Yes. The auth feature is wired to Supabase out of the box (Apple, Google, Email/Password, Forgot Password). Tokens are stored in expo-secure-store. Swap in Firebase Auth or your own API by replacing one client.
How do I get access?
We're in Beta. Submit your name, email, and GitHub username on this page. Access to the GitHub repository lands in your inbox within 24 to 48 hours.