Catching Status Notes That Had Gone Stale
My notes and jobs kept claiming things that were no longer true. I built a nightly check that finds where two files disagree or a promise has lapsed, and learned that most of my old checks had never really been checking anything.

Key Results
- 813 note files scanned, 9 status contradictions and 20 late promises found
- Open drift items went from 46 to 20 in 3 days
- 12 of 20 broken-link flags proved false on a second look
- A date check that had never fired raised 25 flags on its first working run
- 10 of 11 open decisions cleared in one reply
What was broken
I run my work on a large set of notes and small automated jobs. By August 2026 I had a feeling that some of the notes were lying to me, though never on purpose: each one was true when it was written, and nobody updated it after things moved on.
So I measured it. I ran a read-only scan over 813 note files in two passes. It found 6,826 lines that stated a status, grouped them into 754 items and checked which of those items disagreed with each other, and 9 real contradictions survived. It also found 3,358 lines with dates, of which 477 were past due, and after I checked each of those for a later fix, 20 promises were still open and late.
The surprise was the kind of fault I found. Nothing was lost and nothing was hard to find; the files were well linked and simply wrong.
Four fixes, for example, had been applied but still sat in the waiting folder, and the notes around them had said "not applied" for 52 to 72 days. My own dashboard said one helper had been switched off for 74 days, while a card at the top of the same page said none were off. One job had run every week since early June, while its own instruction file said in capitals that it wasn't switched on, and a future session reading that line could easily have stopped a working job.
The checks I already had caught none of this. One file-health job reported zero problems on 8 runs in a row, and it wasn't broken; it was answering a question nobody needed answered. A weekly link report arrived every Sunday and had been acted on zero times in ten weeks, and four of its logs were identical byte for byte. One date check had never fired in its whole life, because it crashed on every date it found and the crash was hidden.
What I built
I built a nightly drift watch with a narrow job: find two files that disagree, or a promise that has lapsed, and put each one in a queue.
The watch may fix things on its own only where there's one correct answer and an easy undo. I allowed four such cases, such as a fix that has been applied but is still listed as waiting. Everything else it proposes to me, a few items at a time, each with its own recommendation.
If I ignore an item for three nights running, the watch parks it and brings it back a month later. The queue started with 46 active items.
The first armed run failed, because I'd set the watch to run somewhere that couldn't reach my files, and it would have failed without a sound every night. I moved it to my own machine, and the fifth run worked end to end.
I also fixed the date check that had never fired. Its first working run raised 25 flags, and one of them was a review I'd set for myself that was already overdue.
What changed
By the eighth run, three days after I armed the watch, open items were down from 46 to 20. That number needs an honest footnote: 6 items of the drop were parked after three nights with no decision from me, and parking an item leaves it unsolved.
The biggest speed-up was in how I made decisions. One night I was tired and had 11 open calls, and I received them as one table showing the item, the question, the recommendation and one line of reasoning. I replied "all yes except one", and 10 of the 11 were cleared in a single message.
Checking also caught mistakes in the checker. Of 20 broken-link flags, 12 had never been real. In 8 of those, a note said "move this file here later", and the tool had read that future address as a broken link, so an automatic fix would have rewritten those notes to say "move this file to itself".
And I got one thing badly wrong. The watch told me a weekly client update had never been sent, because the drafts were still sitting in the folder. The update had in fact gone out every week from my other mailbox, which no tool could see. The real finding was that blind spot: nothing in my system could confirm that a message was sent, only that a draft existed.
What I would do differently
I'd test every check once on something I know is broken, because a check that has reported zero for its whole life deserves suspicion rather than trust. I'd also close the status note in the same sitting as the work, so that applying a fix and updating the note that calls it pending happen together.
I'd treat a missing record, in the one place a tool can see, as a lead to follow up and never as a finding. And I wouldn't let a falling count fool me again; I'd ask how many items were actually decided and how many had simply timed out.
Want a similar result?
Let's talk about installing the right operating system for your business.
Book a Call