Building one complete app teaches more than ten tutorials. Pick one capstone, follow the weekly plan, use the prompts from each chapter, and ship it.
Core concepts
- Project 1: Habit tracker (local-first): Expo Router tabs, NativeWind UI kit, SQLite storage, Zustand filters, local notifications and an offline-first streak calculation. No backend required. — Expo SQLite
- Project 2: AI notes with Supabase: Email auth, notes table with RLS, TanStack Query, an edge function that summarises a note with per-user limits, and Maestro sign-in test. — Supabase Expo quickstart
- Project 3: Subscription app: A niche tool with a free tier and a Pro entitlement through RevenueCat, push reminders, Sentry and a staged release with EAS Update. — RevenueCat docs
- Weekly rhythm: Week 1 spec, rules and UI kit; week 2 screens and state; week 3 data, auth and security; week 4 tests, release and launch. — Expo docs
- Definition of done: Runs on two real devices, tests for critical paths, no secrets in the bundle, store assets ready, privacy policy and account deletion in place. — App Store Review Guidelines
- Launch and feedback: List the app on relevant directories, share it in communities, collect feedback in-app, and ship small OTA improvements weekly. — SaaS directories
Concept videos
YouTube channels and searches for these topics. On a large screen they open in the reading panel next to the roadmap.
- Build a full Expo app — YouTube search: build full mobile app Expo Supabase from scratch
- Indie app launch — YouTube search: launch indie mobile app App Store first users
- notJust.dev projects — notJust.dev channel
Explanation
Pick the smallest useful app
Choose one user, one job and three screens. Resist features until the core flow works on a real phone. Write out-of-scope items in SPEC.md and give that list to the agent.
Four-week plan
Week 1: spec, AGENTS.md, UI kit, navigation. Week 2: screens, state and storage. Week 3: backend, auth, security and AI or payments. Week 4: tests, QA on devices, store assets, beta, submit.
Marketing starts early
Study ranked apps in the Top Mobile Apps database, write your store listing in week 2, and collect an email list or beta testers before the build is done.
What Is It?
A capstone is a complete project that applies every chapter of this roadmap from spec to store.
Why Should I Care?
Shipping end-to-end exposes the gaps tutorials hide, and a live app is the best portfolio and the start of real feedback.
Real World Analogy
It is the final exam of a cooking course: you plan the menu, shop, cook for guests, and serve, not just practise knife skills.
Visual Mental Model
Week 1 Spec + UI kit -> Week 2 Screens + state -> Week 3 Backend + security -> Week 4 Tests + releaseBeginner Explanation
Choose Project 1, write the spec in an hour, and follow the weekly plan with the prompts from each chapter.
Smallest Possible Example
SPEC.md (habit tracker)
User: people building one daily habit.
Screens: Today list, Add habit, Stats.
Data: habits (id, name, schedule), checkins (habitId, date).
Out of scope: social features, accounts, cloud sync.
Risky assumption: users want streaks more than reminders.Common Mistakes
❌ Starting with a huge idea.
You never finish; pick three screens and ship.
❌ Leaving store assets to the last day.
Screenshots, policy and listing copy take longer than expected.
❌ Not testing on real devices until launch week.
Real-device bugs need time to fix.
Real Project Example
Many successful indie apps started as a narrow tool for one audience, shipped small and improved weekly through updates.
Build This Mini Project
Ship Project 1 to TestFlight and Google Play internal testing with five testers and collect their feedback.
Interview Question
How do you decide what to cut from a first release?
Keep the one flow that delivers the core value, list everything else as out of scope in the spec, and validate with real users before adding features.
Key Takeaways
- Pick one project and finish it.
- Four weeks: spec, screens, data, release.
- Define done before you start.
- Plan store assets and marketing early.
Stack for this step
| Layer | Pick | Why |
|---|---|---|
| Project 1 | Expo + SQLite + Zustand | Local-first, no backend. |
| Project 2 | Expo + Supabase + edge function | Auth, RLS, AI. |
| Project 3 | Expo + RevenueCat + Sentry | Monetised and monitored. |
Prompts for Cursor, Claude Code & Codex
Paste these into Cursor, Claude Code, Codex, or any coding agent. Replace the bracketed parts with your app details, and keep the agent in plan mode first so you review the approach before it edits files.
Four-week plan
From SPEC.md, create a four-week build plan with daily goals, the prompts to run, and a demo you can show at the end of each week. Highlight the riskiest item and schedule it first.Definition of done audit
Audit the repo against this definition of done: runs on two devices, tests for critical paths, no secrets in the bundle, privacy policy and account deletion present, store assets ready. Return pass/fail with next actions.Launch kit
Write a launch kit: store listing copy, a short Show-and-tell post for communities, ten directories to list the app on with reasons, and five questions to ask early users.Claude Code skill for this step
Save this as .claude/skills/weekly-review/SKILL.md in your project so Claude Code loads it automatically when the task matches. The same text works as a Cursor rule or an AGENTS.md section.
---
name: weekly-review
description: Use at the end of each build week to check progress against the plan and update the spec.
---
1. Compare shipped features with the week's goal.
2. List bugs found on real devices and turn them into tests.
3. Update SPEC.md: cut or defer anything slipping.
4. Plan next week's top three tasks.Rules, databases & blogs to use
- SaaS directories: list your app on 390+ directories
- Directory launch playbook: directory submission strategy
- Micro SaaS ideas: ideas that make money
- Top mobile apps: market research
- YC companies: see which problems startups fund
- Investors database: if you plan to raise
- SaaS starter kit lessons: starter kit approach
Resources
The 60-second story
The final exam of a cooking course is not another knife drill. You plan a menu, shop, cook for real guests and serve before they get hungry. You learn what the class never showed: timing, plating, and what to do when the oven is late.
A capstone app is that dinner. You write the spec, set the tokens, build screens, connect the data, secure the doors, test on real phones, and put it in front of real users. By week four you have a live app and a list of what to cook next.
Mental model: cook the whole dinner once.
