Integrate weather forecast data into a trip planning agent

Steps

1. Choose a weather API suited to travel use cases: Open-Meteo (free, open-source, no key required for basic use), OpenWeatherMap, Tomorrow.io, or Weatherbit — each offers different forecast horizons (5 to 16 days) and historical data access.
2. Query by coordinates (lat/lng) rather than city name for precision — resolve destination city or airport to coordinates using the Google Geocoding API or a local airport coordinate dataset first.
3. For trip planning, request daily aggregates (min/max temperature, precipitation_sum, wind_speed_max, weather_code) rather than hourly data to reduce response size; most APIs support a daily parameter or aggregation mode.
4. Map weather codes to human-readable conditions and travel advisories: heavy rain (codes indicating thunderstorms or heavy precipitation) should trigger packing suggestions or activity warnings, not just a raw number.
5. Limit forecast confidence communication: beyond 7 days, weather forecasts have high uncertainty — label forecasts for days 8–14 as 'outlook only' and do not make irreversible booking decisions (e.g. activity bookings) based on them.
6. Cache forecast responses with a TTL of 1–3 hours — weather APIs charge per request and forecasts do not change meaningfully minute-to-minute; bust the cache if the user explicitly requests a refresh.