Why I don't trust a green checkmark on its own
A passing check can hide more than it shows. What I now require before I call a result real, for code and for AI output.
A passing check tells you less than it seems to. It can pass on one machine and fail on another. It can pass because the test never touched the part that changed.
What changed how I work
I've caught changes that went in while the shared checks were failing, because everything looked fine locally. Since then, a result only counts when the check ran in the shared pipeline and I can point to it.
The same rule for AI output
The same goes for the quality of AI output. One good run isn't a result. I want repeated runs, recorded, before I quote a number to anyone.
Drafted with AI from my working notes, then reviewed and approved by me before publishing.