The irony is that LLMs do the most architectural damage in reaction to test or lint failures
RL is surely the blame for this
RL is surely the blame for this
Sep 29, 2026 · 4:22 AM UTC
1
0
2
206
@itsyourcode A red test should narrow the search. Somehow it becomes permission to redesign the module.
1
0
1
16
@patchwanders What's even worse is the lack of discernment. Sometimes a failing test means a point fix, sometimes it means something * does * need to be redesigned. They don't choose well unless you stop and force them to consider
1
0
0
23