I can see Claude's reasoning

Claude suddenly started to show the whole reasoning process when given a command, literally pages and pages of thoughts on how he perceives various elements of my prompts and how he decides what to do. I asked about it and my request was blocked [reasoning extraction]. I shrugged, I get it, obviously. But then so did every single after that - prompts which had nothing to do with it. Switching to different model didn't do much - same reaction. Can't even ask him to prepare handoff files before /clear. Anyone had that happen?