Soil Moisture Sensor Triggered Scheduling via IFTTT

First time Rachio user here, just installed a gen 2 a few weeks ago. So far I’m very impressed and wish I would’ve made the jump to a smart controller a long time ago. Though I’ve noticed that it’s probably over-watering more than I would’ve on my old dumb controller. Water is fairly expensive here so I tried to adjust the zone settings to be more thrifty with the flex daily schedule, even if it results in some areas of dormancy in my lawn.

At any rate, I was thinking more about the predictive nature of the scheduling algorithm and thought that a real-time soil moisture sensor would improve the overall efficiency as it would add a ground-truth data point. However, after some searching on these forums, it doesn’t seem to be possible to manually correct the soil moisture values directly, so they are whatever rachio thinks they are on the best available data that is exposed to the app. Which I’m fine with, since they do at least allow some flexibility in tweaking the schedule via IFTTT, albeit in a more indirect way.

Long post, longer, I came here to ask if anyone was thinking of or has already implemented a similar system of automatically triggering stops, skips, or restarts via IFTTT based on external data such as soil moisture readings or anything else from an unlisted PWS. If so, how did you set it up, and what specific IFTTT triggers/actions did you come up with?

My thoughts so far are to just add a couple simple automations which would be to stop the flex daily schedule if soil moisture is above a certain value (e.g. 70%), and then restart it if it drops below a certain value (e.g. 60%). Obviously this will have some limitations since the soil moisture readings may still be off per the actual readings and I might eventually have to attempt to correct this via the “Empty/Fill” functions for each zone. I do realize that the soil moisture percentages used by Rachio are not quite the same as the soil moisture percentages as reported by a soil moisture sensor, but I hope to use it just for some minor adjustments relative to the predictive values assumed. Just curious if anyone else had any other ideas beyond this.