Passing the JLPT N2 has been an itch I've wanted to scratch for years. I had the vocabulary foundation, the daily immersion, the motivation — but not a study tool I'd actually want to open every morning. So I set out to have one built: a spaced-repetition flashcard app for N2 vocabulary, designed with the same care I'd bring to a client project. I provided the concept, the content direction and the design inputs; Claude, Anthropic's AI, wrote every line of code across a series of iterative sessions.
The result is a React and TypeScript PWA wrapped for iOS via Capacitor, with cloud sync, progress visualisations, and over 6,000 vocabulary cards across 105 decks. Before any UI code was written, I developed a full design system to ensure consistency across every screen. The spaced-repetition engine uses a custom implementation of the SM-2 algorithm, and the animated kanji stroke order data comes from KanjiVG (CC BY-SA 3.0).