Putting Every Content Rule Behind a Check
My content system had written rules that nothing enforced. In one day I gave every one of them a weekly check, and the checks found problems that the rules on their own never had.

Key Results
- Content rules with a working check went from 16 of 29 to 28 of 28 in 1 day
- The weekly check page grew from about 15 rows to 48 rows
- 79 files read and judged one by one; only 27 held a real content rule
- 1 false alarm that had fired on 57 days out of 84 was narrowed to a clean run
What was broken
My content system runs on rules about what gets written, how long it is, where it goes and what must never be said. The rules live in documents, and the AI agents that write the content read some of those documents before they start.
In August I asked a plain question: which of these rules is actually in force? Nobody knew. There were a lot of rule documents and a weekly check page with about 15 rows, and nothing linked any rule to the check that was supposed to enforce it.
Then I found a clear case. On one day in July I had told the system, in writing, to stop posting to one social channel, and the note was sitting on disk. The schedule kept asking for that channel for 45 more days anyway, which added up to 15 slots due and 0 posts. That one unenforced note was the single biggest reason my posting numbers looked bad.
What I built
First I had to find the real rules, because a folder full of files doesn't tell anyone which of them are rules. So I opened 79 files and judged each one against a simple test: does it change what gets written, how, where, or whether it may go out at all? Each file got a verdict and a written reason. 21 governed content, 20 sat next to content without governing it, and 38 had nothing to do with it.
Across that pass and an earlier one, 27 documents stated a real content rule. I loaded all 27 into one database and then checked each rule for a gate, and 13 of them had nothing enforcing them at all.
Then I closed them, one rule at a time. For each rule I wrote a small check that runs every week and reports a count, so instead of a label like "fine" the page shows something like 46 of 46. Every check follows three habits:
- It says when it can't look. A check that couldn't open a folder must never report zero and call that a pass.
- It ships with a test that plants a fake mistake and makes sure the check catches it. Three checks passed every row on their first run, and the planted test showed that one of them was wrong: it would have failed an honest price list forever.
- It never skips anything without saying so. When a rule is retired it leaves the count, and the page says how many rules left and why.
What changed
By the end of the day, 28 of 28 rules had a working check, against 16 of 29 that morning, and the weekly page had grown from about 15 rows to 48. The checks also found things that the rules on their own had hidden.
One live job had been pointing at a file that hadn't existed for two months. The main content job sorted its style files by name, and when I tested it with versions 1.9 and 1.10 both present, the name sort picked 1.9, which means that on the day a tenth version was signed, the job would have kept running the old one without a word to anyone.
A safety alarm had fired on 57 days out of 84. Every time, it had caught the same ordinary word and never a real problem. Because it always fired, a real alarm would have landed next to one that everyone had learned to ignore. I narrowed it to match that one term as a whole word only, and it ran clean for the first time since June.
Some of the bad news turned out to be bad counting. My tracking sheets said three channels were at about 70 percent of the posting target, while the live channels showed about 90 percent, which is the target. The sheets were missing posts, so that check now reports "not measurable" instead of "fail" until the sheets are fixed.
Six rows still fail, and they should. Each one is a real gap that now has a number next to it.
What I would do differently
I'd link every rule to its check on the day I write the rule, because a rule with no check is only a note. I'd also stop trusting any status line written inside a rule document, since three times in one day a document said something wasn't built when it was already live.
I'd sort anything that carries a version number by version, never by name. And I'd treat every check that passes on its first run as a check that still needs testing, rather than as a result I can trust.
Want a similar result?
Let's talk about installing the right operating system for your business.
Book a Call