miTi is an open-source Nostr client for finding and publishing calendar events. It speaks NIP-52, so calendars and events created anywhere on the network appear here, and anything you publish from miTi propagates back out to every other Nostr client. Everything runs in the browser: discover what's happening, open the full details, place an event on a map, and add your own to the network. miTi also learns what you like — right on your device — to push the events worth your time to the top, and can sync your settings and feedback privately across your devices.
Try it: https://www.letsmiti.app
- 📅 Browse NIP-52 calendars and date/time-based events (kinds 31922–31924)
- 🔎 Detailed event and calendar pages, including NIP-52 RSVPs (kind 31925)
- ❤️ Personal recommendations — like, dislike, report, or hide events; miTi learns your taste and reorders the feed, plus a Suggested page and a My feedback page. It all happens on your device.
- 🚫 Built-in spam filter that hides junk and low-effort events from the feed and map
- 🔄 Optional cross-device sync of your settings and feedback, end-to-end encrypted to yourself over Nostr (NIP-78 + NIP-44)
- 🛰️ Bring your own relays and Blossom upload server
- 🗺️ Map view with a distance-radius filter and geolocation, plus a reusable location filter for the feed
- ✍️ Publish your own events and calendars, with Blossom cover-image uploads
- 🌍 Timezone-correct scheduling and display
- 📥 Export any calendar to
.ics, or subscribe to it over webcal - 🔗 Auto-generated OpenGraph previews so shared links unfurl cleanly
- 📲 Installable as an app (PWA) and usable offline
- 🌐 Available in English, German, and Spanish
When you like, dislike, report, hide, or RSVP to an event, miTi turns the words in that event (its title, tags, and — if you opt in — its description, summary, and location) into a small private profile of what you tend to like. It then ranks every event by how well it matches, lifts good matches to the top of the feed, fills the Suggested page, and pushes likely junk to the Spam page.
All of this runs in your browser and is stored only on your device (in IndexedDB). Nothing about your taste is ever sent anywhere in the clear. If you turn on sync, your settings and feedback are encrypted to yourself (NIP-44) and saved as Nostr app-data events (NIP-78, kind 30078), so only you can read them and only your own devices pick them up. Sync needs a signer that supports NIP-44 (a browser extension, a remote signer/bunker, or a local key); read-only logins simply skip it.
- Framework: Next.js 15 (App Router), React 19, TypeScript 5
- UI: Tailwind CSS v4 + shadcn/ui + Base UI + lucide-react icons + Geist font
- Nostr:
@nostr-dev-kit/ndkv2 withndk-cache-dexie(IndexedDB), nostr-tools, nostr-login - Personalization: on-device taste engine in a Web Worker, stored in IndexedDB (Dexie); private cross-device sync via NIP-78 (kind 30078) + NIP-44 encryption
- PWA: Serwist (
@serwist/next) — installable app + offline support - Data fetching: TanStack React Query v5
- Maps: react-leaflet v5 + Leaflet, CARTO basemaps
- Geocoding: Nominatim (OpenStreetMap) + Overpass, cached in IndexedDB via Dexie
- Uploads: Blossom protocol (default
blossom.nostr.build, configurable in settings) - i18n: i18next + react-i18next
Nearly all of the work happens client-side: NDK opens WebSockets straight from the browser to the relays. The only server-side code is a thin compatibility layer (the OpenGraph image routes, generateMetadata, and the ICS feed) that lets non-browser consumers such as link-preview crawlers and calendar apps still get useful output.
- Node.js 22 or newer
npm install
cp .env.example .env.local # optional; every variable has a sensible default
npm run devThen open http://localhost:3000.
| Command | Description |
|---|---|
npm run dev |
Start the development server (http://localhost:3000) |
npm run build |
Production build (runs lint first via prebuild) |
npm run start |
Start the production server |
npm run lint |
ESLint, with zero warnings allowed |
npm run lint:fix |
Auto-fix lint issues |
Every variable is optional. With nothing set, the app runs on localhost:3000 in dev, and on a deploy it auto-detects its public origin from the request host, so a free *.vercel.app URL works with zero configuration. Set these only when you want to override that behavior.
Copy .env.example to .env.local to get started. Anything read by the browser must carry the NEXT_PUBLIC_ prefix, which Next.js inlines into the client bundle at build time.
| Variable | Default | Purpose |
|---|---|---|
NEXT_PUBLIC_SITE_URL |
auto-detected | Canonical public origin (scheme + host, no trailing slash) used for canonical links, OpenGraph/Twitter URLs, and the metadata base. Set it once you have a stable custom domain so SEO canonicals don't drift between *.vercel.app and the real domain. |
NEXT_PUBLIC_PLAUSIBLE_DOMAIN |
unset (analytics off) | The site key registered in your Plausible dashboard. The analytics <script> loads only when this is set, so localhost and preview deploys report nothing. |
NEXT_PUBLIC_I18N_DEBUG |
false |
Set to "true" for verbose i18next logging in the console. |
The runtime and host provide these automatically; they feed the base-URL resolver in src/lib/baseUrl.ts as fallbacks.
| Variable | Source |
|---|---|
NODE_ENV |
Set by Next.js (development / production) |
PORT |
Dev/start port (defaults to 3000) |
VERCEL_PROJECT_PRODUCTION_URL |
Stable production domain, injected on Vercel |
VERCEL_URL |
Per-deploy URL, injected on Vercel |
The resolver picks the first of these that is available, in priority order: NEXT_PUBLIC_SITE_URL, then the x-forwarded-host / x-forwarded-proto request headers, then the Vercel variables, and finally localhost. That chain keeps dev, free deploys, and a future custom domain working without any code changes.
The production instance runs on port 4000 under PM2 (configured in ecosystem.config.cjs). It also deploys to Vercel as-is, where the host and base URL are detected for you.
Pull requests and issues are both welcome, whether it's a bug, a feature idea, a translation, or a fix. For anything larger, opening an issue first to talk through the approach tends to save everyone time.
miTi is distributed under the GNU Affero General Public License v3.0 (AGPL-3.0). You may run, study, modify, and share it; in exchange, any networked deployment or derivative must stay AGPL-3.0 and make its own source available. The full terms are in LICENSE.
Shipped without warranty, so run it at your own risk, and built in the hope of a freer, more decentralized web.