Weather Life Voice Broadcast

Problem

Weather data is easy to fetch, but accessibility and voice workflows need structured weather converted into listenable copy plus a clear illustration. A raw weather API usually returns temperature, wind, and precipitation, while leaving city-name disambiguation, date completion, lifestyle advice, TTS, and image generation disconnected.

How It Works

Provide required city and optional date in YYYY-MM-DD; if omitted, the current date is used. The skill runs an async pipeline:
- Parse and validate: extract city and date, filling in missing dates.
- Disambiguate city: use an LLM to confirm the target place and reduce mismatches from aliases or colloquial names.
- Query weather: call ShowAPI for current or forecast conditions and indices.
- Build copy: generate clothing and travel advice, then create an image prompt.
- Generate media: start TTS and text-to-image tasks in parallel, polling task_id until audio and image finish.
- Return bundle: provide broadcast copy, audio URL, and image URL for HTML or players.

Notes

Real calls require showapi_appkey; without it, only a mock preview is returned and the async task scripts are not executed. All tasks are async, so synchronous mode is not supported; feedback should be provided after each polling round to avoid a silent wait.