Lohnt sich dein eigenes Wanderfugl?

Build a local-first trip planner web app called Fugl. No accounts, no cloud, no telemetry.

Stack, non negotiable:
- Next.js with the App Router and TypeScript
- SQLite via better-sqlite3, file at ./data/fugl.db, created and migrated on first run
- Tailwind for styling, no component library
- Leaflet with OpenStreetMap tiles for the map
- Runs with: npm install, npm run dev

Data model:
- Trip: id, name, destination label, start date, end date, notes, home timezone
- Item: id, trip id, kind (flight, train, lodging, activity, food, note), title, start datetime, end datetime nullable, timezone, location name, lat nullable, lng nullable, confirmation code, url, cost, currency, notes
- Packing item: id, trip id, label, packed boolean

Features in scope:
- Trip list page, create and edit trips
- Trip detail page grouped by day, each day showing items sorted by start time, with kind icons and a running per-day cost total
- Add and edit items in a side drawer, with lat and lng entered manually or pasted from a maps URL
- Map panel showing all items with coordinates for the trip, pins numbered by day
- Packing list with checkboxes, persisted
- Export trip to .ics with one VEVENT per item, timezone aware
- Printable itinerary view at /trips/[id]/print, clean single column, no nav chrome
- JSON export and import of a whole trip
- Full text search across item titles, notes and confirmation codes

Explicitly out of scope:
- Flight, hotel or price lookups of any kind
- Email parsing, OAuth, sharing links, multi user
- Push notifications and native apps

Use .env for the optional map tile URL and nothing else. Seed one demo trip with three days of items so the UI is not empty. Write a README with run steps and a one line note that all data stays in ./data.