Mali-P/Levant-Lab

# Levantry Hebrew + Levantine Arabic flashcards, at levantry.app. A private, offline-first, mobile-first flashcard app for drilling **Hebrew and Levantine Arabic at the same time**. A card only counts as correct when both languages are recalled correctly. Built as an installable PWA: React + TypeScript + Vite, IndexedDB via Dexie, Zustand for state, Framer Motion for the card gestures. ## Running it ```bash npm install npm run dev # development server npm run build # production build into dist/ npm run preview # serve the production build npm test # unit tests for the study engine and validation npm run serve # build, then serve the app and the sync server together ``` ### Putting it on your phone `npm run build`, serve `dist/` over HTTPS (or `localhost`), open it in Chrome on Android and choose **Add to home screen**. After the first load it works with no network: cards, progress, statistics and device voices are all local. ## Syncing the phone with the laptop The app is still local-first: every device studies from its own IndexedDB and works with no network at all. Sync is an optional extra that lets two devices agree, and it runs on your own machine — there is no cloud account. On the laptop: ```bash npm run serve ``` That builds the app and starts the sync server, which serves both the app and the API from one address. It prints a token and the addresses it is reachable on: ```text token 3f9c… Say the following in everyday spoken Palestinian Levantine Arabic, in the > accent of a native speaker from Jerusalem… That instruction is `GEMINI_ARABIC_STYLE`, and it counts as part of the voice: editing it re-records every Arabic clip. A Modern Standard Arabic voice is never asked for, and the vocabulary itself is never rewritten towards MSA. The Azure path is shelved rather than removed — `ARABIC_TTS_PROVIDER=azure` brings it back, which is how the two accents get compared. Every voice is set through the en …