Keep a short incident log when something breaks

Every time an automation breaks in a real way, write down what happened, what caused it, and how you fixed it. Keep these entries together in one place instead of letting the memory of each incident fade after the fire is out.

This matters because the same kind of failure tends to repeat in slightly different forms, and a written incident log means you recognize the pattern the second time instead of solving the same problem from scratch again. It also helps anyone new to the automation understand its history quickly, instead of hearing secondhand stories about "that one time it broke" with none of the useful detail attached. I keep my own log in one place, and I've noticed the same kind of failure tends to show up again in a slightly different costume.

Keep each entry short: what broke, what caused it, what fixed it, and anything you would do differently.

Try this today: write your first incident log entry for the last time any automation gave you trouble, even if it happened a while ago.