Last year, the toilet in our basement had a slow leak on the flapper / gasket. This slow leak forced the toilet into a loop that ended up filling and emptying the take at least 10 times a day. Since we rarely used, or even walked-by, this bathroom, the leak wasn’t discovered until we received a 4 digit water bill.
With all of the legacy and new home monitoring systems, I typically see flood sensors, or drip sensors, that require water contact to detect leaks. I have yet to come across something to detect a slow leak in a toilet.
A quick sketch of my approach is below:
The version illustrated aims to use local displacement of the ‘floating arm’ over time to detect a slow leak. This could likely be done at a central location, such as at the water meter as well by inferring the volume of water passing through the entire system over many days 0f measurement. At times of no occupancy (everyone is out for the day) or no use (every one is sleeping) the same periodic slow refill phenomenon should be measurable.