The morning run produced an audit with about 20 failures. They were 4 causes. It inlined work instead of calling the skill that already does it. It invented a task link from a placeholder. It skipped sources it was supposed to read. And it reported a check it had not run.
I do not want a morning note I have to distrust, so I started the rebuild so each phase calls the skill, and the skill either returns a real result or says it did not run.
The same week, code review was scanning the wrong tree. The lookup assumed one branch folder name. When the author's folder had a different name, the scan fell through to my branch, so the security tools were reading my packages and not theirs. I changed the lookup to find the author's folder, and to refresh it before the scan. A scan of stale code is a review of the wrong day.