Involves and made only by, are 2 different things.
I use agentic coding in my daily work. I do make mental model of the code I write and I also test the code, exactly the same way, as when written completely manually.
I understand it can be difficult to label, and there's an inconveniently large grey area here, but there is a difference between plain vibe-coded software and software built with AI under the control, direction and validation of a developer.
The distinction is probably not very important for small applications, as nobody cares if a minor script or a one-shot data processor has been vibe-coded, but for large applications it surely matters in the long term.