Global City Weather Forecast

Problem

When users ask about weather in natural language, city, time frame, and required fields can be mixed together. The skill needs to identify the target city and turn sparse API data into a readable summary.

How It Works

• Trigger and parameters: Recognizes requests like “Shenzhen weather” or “weather in Beijing”; falls back to Shenzhen when no city is given.
• Data source: Calls the free wttr.in service without an API key, supporting Chinese, English, and pinyin city names.
• Parsing: Reads current_condition fields such as temp_C, FeelsLikeC, humidity, windspeedKmph, winddir16Point, and uvIndex, then parses today’s and next two days’ maxtempC, mintempC, and hourly entries.
• Output: Starts with a short current-condition summary, adds a table for temperature, feels-like, humidity, wind, and UV, then lists a 3-day forecast with practical tips such as bringing an umbrella or applying sunscreen.

Boundaries

Best for quick lookups and lightweight guidance, not minute-by-minute warnings. If wttr.in times out, check the network; if a city is not found, try its English name. Descriptions default to Chinese but can switch to English when the user asks in English.