I use success loops and longer-running prompts to reduce tinkering, and I prefer planning and verifying code before building over a review mode that feels too token-intensive.
I observed that YOLO had become the default mode on command code.
I made this observation on 2026-11-13 07:26 UTC.
My prompts now run for at least 15 mins each.
My prompts save a lot of tinkering time.
Review mode feels too aggressive on tokens to me, since it fights after writing the code.
I feel that verification and planning the code before building seems better.