112 Scheduled Jobs, and How Many Really Ran
I had 112 scheduled jobs quietly running my operating system. When I checked, only 66 showed any proof of ever running. So I built a record that makes every job show its own work.

Key Results
- 66 of 112 scheduled jobs showed any proof of running; 3 had never appeared at all
- 24 of 25 one-time jobs were still on the list, up to 94 days past their date
- 61 of 112 jobs showed a run in the last 7 days
- Jobs that record their own runs went from 1 to 6 of 112 in 2 days
- 1 hidden group of 10 more jobs found, taking the true total to 122
What was broken
My operating system runs on scheduled jobs. Each job is a written set of steps that an AI agent carries out on a timer: some write content, some check my websites and some remind me of things.
In August I went looking for one important job and found a whole folder of them instead. It held 112 jobs. They weren't in my computer's own scheduler or in my cloud task list, and no audit had ever opened that folder.
So I asked a simple question: of these 112 jobs, how many really run?
The answer was worse than I'd thought. Only 66 showed any proof of running. 24 had been seen just once, in a single session that listed them, and never in an actual run. 3 had never appeared anywhere at all, and the rest either ran rarely or had gone cold months ago.
There was a second problem. 25 of the jobs were one-time jobs, meant to fire on a set date and then finish, and 24 of them were still on the list. The oldest was 94 days past its date.
On top of that, almost no job wrote down that it had run. On the first day only 1 of 112 did, so the list told me what I'd planned and nothing about what had actually happened.
What I built
I built a small record in a single database file. It holds every job, the group it belongs to, and whether there's proof it ran.
The main rule is that every check has to give a number instead of a label. A check that says "fine" or "broken" tells me nothing next week, while "61 of 112" lets me see whether the figure moves from one week to the next.
I added one tiny helper that any job can call at the end of its run. It writes a single line saying that this job ran at this time, and it needs no setup.
Then I set up a weekly check that counts again from the files themselves and sends me one line if anything has got worse.
While building this, I kept tripping over the same mistake. Five times in one day, a check that couldn't look reported that it had looked and found nothing. A check that couldn't open a folder reported zero out of zero and marked it as a pass, and at one point 11 of 15 checks could do this. I fixed each one, so a check that can't see now says plainly that it can't see.
I also found that my first weekly check had never once worked on its schedule. It ran in the cloud, and a cloud job can't reach the files on my computer, so every week it fired, did nothing and still looked fine.
That led to one more find: there were 10 more jobs in the cloud, outside the folder, which made the true total 122 instead of 112. I added them to the record, but I count them separately, because they can never write to my computer and counting them in would make that number fail forever.
What changed
In 2 days, the number of jobs that record their own runs went from 1 to 6. That's still small, but it's now a real number I can watch climb.
61 of 112 jobs showed a run in the past 7 days, and for the first time I know which ones didn't.
I didn't switch anything off. I ruled that we build the record first and fix things after it exists, because stopping five problems without a record would only hide the next five.
What I would do differently
I'd make every job write one line when it runs, starting from the very first job. It costs almost nothing up front, and adding it later to 112 jobs costs a lot.
I'd give every one-time job an end date. A job that should have finished and didn't is a small risk that grows quietly.
I'd treat any sentence with "never" or "zero" in it as a claim to test before I believe it, because mine turned out to be wrong more than once.
And I'd match jobs by their fixed ID instead of their name. When one job was renamed, my count showed 123 instead of 122 for a while. The job hadn't doubled; my counting had counted it twice.
Want a similar result?
Let's talk about installing the right operating system for your business.
Book a Call