Plan mode does have different permission controls though? Its difficult for me to parse the docs on this but it does seem to imply more than just a prompt although less than a full sandbox. https://code.claude.com/docs/en/permission-modes#analyze-bef...
replyThe way it worked originally was it would trigger a different system prompt, and then the permissions system would prompt you (even on YOLO mode) before running any non-readonly actions.
replyCursor would also write a plan document, which was useful when working on larger tasks (due to context size). I still find that part useful today.
early plan mode would just refuse to make edits at all
replyPlan forced the model to write down and outline what it was going to do before writing the code. As Boris said up thread, he created it to prevent it from just rushing into writing code. So that regard, it did act like a permission system.
replyIt helps to think more abstractly when approaching problems like this.