A routine that opens the blinds at 7 a.m. and closes them at 8 p.m. works fine until the first overcast week, when the living room stays dim all morning, or the first heat wave, when the thermostat cools an empty house because the schedule doesn’t know it’s 95 degrees outside. Pairing an existing Home Assistant or Alexa setup with a weather integration fixes that gap: instead of running on the clock alone, your routines can check tomorrow’s forecast and adjust blinds, thermostats, and sprinklers before the weather actually arrives. It takes one added integration and a handful of automation conditions, not a rebuild of what you already have.
Builder · Level 2 of 3
Why fixed schedules fall short
A time-based routine is really a bet that today looks like an average day. Most days it does, which is exactly why the failure shows up as an annoyance rather than a crisis: the sprinklers run the morning after it rained, the blinds stay open through a west-facing afternoon that hits 100 degrees, or the thermostat holds a daytime setback on a day cold enough that the house never recovers by evening. None of that breaks anything. It just means the automation quietly stops being useful, and most people notice the pattern after a week or two of small mismatches rather than any single dramatic failure.
The fix is not a smarter clock. It’s giving the routine one more input: a forecast for today or tomorrow, pulled from a weather integration your hub already supports, checked as a condition before the automation runs.
Weather integration options for Home Assistant
If you completed Home Assistant’s initial onboarding, you likely already have a working weather entity: Home Assistant defaults new installs to the Meteorologisk institutt (Met.no) integration for your home’s location, and it needs no API key or account. It polls roughly every 55 to 65 minutes and covers current conditions plus multi-day forecasts, which is enough for most blinds/thermostat/sprinkler routines. Two other official integrations cover cases where Met.no isn’t detailed enough for your area or use case.
| Integration | API key required | Forecast depth | Best for |
|---|---|---|---|
| Met.no | No | Multi-day, updated hourly-ish | Default choice, zero setup |
| OpenWeatherMap | Yes (free tier) | Hourly for 48 hours, daily for 8 days | Short-term precipitation and temperature detail |
| Pirate Weather | Yes (free tier) | Hourly and daily, Dark Sky-style data | Migrating old Dark Sky automations |
OpenWeatherMap’s official integration uses its One Call API, with a free tier that covers 1,000 calls a day, current conditions, an hourly forecast out to 48 hours, and a daily forecast out to 8 days. That hourly window is the detail that matters for a same-day blinds or sprinkler decision. Pirate Weather exists specifically as a drop-in successor to the old Dark Sky integration, so if you have automations that predate Dark Sky’s shutdown, it’s the least disruptive path to keep them running on current data.
One detail worth knowing before you build anything: as of Home Assistant 2024.4, forecast data is no longer sitting in a weather entity’s state attributes. You pull it with the weather.get_forecasts action, targeted at your weather entity, with a response variable that holds the result. Older automation examples that read state_attr('weather.home', 'forecast') directly are for a pre-2024.4 setup and won’t return data on a current install — the step-by-step below uses the current method.
Already running Home Assistant automations for other parts of the house? 20 Home Assistant Automations Worth Setting Up First covers the routines worth building before this one.
Weather integration options for Alexa routines
Alexa’s routine builder is more limited here, and it’s worth knowing that up front rather than hunting for a setting that isn’t there. The native “When this happens” trigger list in the Alexa app does not include raw temperature, humidity, or forecast conditions — you can have Alexa read the weather as a routine action, but you can’t use the weather itself to trigger a routine using only Amazon’s built-in options.
Two workarounds close that gap:
- Enable the Big Sky skill. Amazon’s “Custom Triggers for Routines” developer program lets specific skills add their own trigger categories to the routine builder. Big Sky is one of the skills that onboarded weather-based triggers this way, adding Temperature Change, Wind Speed, and Chance of Precipitation as selectable conditions once the skill is linked. Enable the skill in the Alexa app, then check the “When this happens” menu for a new weather category.
- Bridge through IFTTT and Voice Monkey. IFTTT can watch a weather service for a condition change (Rain, Snow, Cloudy, Clear) and fire a Voice Monkey applet, which can trigger an Alexa announcement or flip a virtual switch. That virtual switch state change is something a normal Alexa Routine can trigger on, which gives you an indirect weather-to-routine path without a skill-specific trigger.
- Confirm your weather entity exists: go to Settings > Devices & services > Entities and search for
weather.. Note the entity ID (typicallyweather.homeorweather.forecast_home). - Create a new automation under Settings > Automations & scenes > Create Automation > Start with an empty automation.
- Set the trigger to a daily time, early enough to act before the day’s weather arrives — 6:30 a.m. works for blinds and sprinklers.
- Add an action step that calls
weather.get_forecastswithtype: daily, targeted at your weather entity, and a response variable namedforecast. - Add a condition template that checks tomorrow’s high temperature or precipitation probability from that response, for example holding the sprinkler routine if
precipitation_probabilityfor the next day is 60 percent or higher. - Add the actual device actions gated behind that condition: skip the irrigation valve’s “run” service call, close the blind covers if the forecast high exceeds your comfort threshold, or nudge the thermostat’s target temperature down before a hot afternoon rather than reacting to it after the house has already heated up.
- Save the automation, then use the three-dot menu’s “Run actions” option to fire it manually once and check the logbook for each entity to confirm it did what you expected, before trusting it to run unattended.
- Forecast misses happen. A forecast is a probability, not a guarantee — a routine gated on “60 percent chance of rain” will occasionally skip watering on a day that stays dry, or water right before an unexpected shower. Set thresholds a little conservative rather than exact, and accept the occasional miss as the tradeoff for skipping the far more common miss of a fixed schedule.
- Weather entity outages. If the integration’s polling fails or an API key issue blocks OpenWeatherMap or Pirate Weather from returning data, a condition template referencing a missing value can throw an error and skip the whole automation silently. Add a fallback branch (or at minimum a notification) for when the forecast call returns nothing, so a bad automation night doesn’t get quietly ignored for a week.
- Old automations built on the pre-2024.4 forecast attribute. These stop returning data after an update and fail without an obvious error message pointing at the cause — if a weather automation that used to work goes quiet, this is the first thing to check.
- Alexa’s IFTTT bridge adds latency. Because it depends on IFTTT’s polling interval rather than an instant push, the Voice Monkey path can lag real conditions by several minutes to occasionally longer — fine for a morning blinds routine, less fine for anything time-critical.
Neither option matches Home Assistant’s native flexibility, so if your smart home already runs through Home Assistant with Alexa layered on top for voice control, building the weather logic in Home Assistant and exposing the result to Alexa as a simple on/off entity is usually less fragile than assembling an IFTTT chain.
Step-by-step: build a forecast-based blinds/thermostat/sprinkler routine
This example uses Home Assistant with the Met.no integration already in place from onboarding — swap in OpenWeatherMap or Pirate Weather by changing the entity ID if you’re using one of those instead.
The condition is doing the actual work here — the trigger just decides when to ask the question. Get the condition template right on one automation, and the same pattern (time trigger, forecast lookup, condition, gated action) covers blinds, sprinklers, and thermostat setbacks with only the threshold numbers changing.
For the Alexa/Big Sky path, the equivalent setup lives entirely in the Alexa app: create a routine, choose the Big Sky weather category under “When this happens,” pick a threshold (for example, chance of precipitation above a set percentage), and choose your smart plug, thermostat, or blind controller as the action. It’s less configurable than Home Assistant’s template conditions, but for a single trigger-to-action pairing it’s faster to set up.
Failure modes and edge cases
Maintenance burden
Once it’s running, this is low-maintenance compared to most Home Assistant automations: Met.no needs no key to expire, and the automation itself only needs attention after a Home Assistant core update changes how forecasts are fetched (as the 2024.4 change did) or if you swap weather providers. If you’re on OpenWeatherMap or Pirate Weather, put a reminder on your calendar to check that the API key is still valid every few months — free tiers occasionally change limits, and an expired or rate-limited key is the quiet cause of an automation that stops updating without any obvious error.
For the Alexa/Big Sky route, maintenance is closer to zero on your end, since the skill vendor operates the trigger logic — the main thing worth checking periodically is whether the skill is still installed and linked after any account or app changes.
Ready to keep building on the same Home Assistant setup? Home Assistant Automations That Cut Your Electric Bill covers the routines worth adding next once the forecast-based basics are working.
Start with one routine — blinds or sprinklers are the easiest to verify since you can watch them react the same day — before expanding the pattern to the rest of the house.