For some additional context, I’m hoping to be able to utilize this tool to automate human-in-the-loop gates. It’s important to be able to ask the user a question, parse their selected responses, and utilize them to record the approvals.
It “works” to have the Agent call some other CLI/Tool after receiving the response, but this also gives the agent the ability to cheat the system by approving the gates without actually getting consent from the human in the loop.
Hey, thanks for the detailed report and the screen recording.
It looks like this is a bug. The docs say preToolUse and postToolUse run for all tools, so AskQuestion should be included too. I’m passing this on to the team.
About the hooks not firing in Plan mode, I’m flagging that separately as well. This might be intentional or not, but either way it should be documented.
I’ll update this thread if I hear anything on either item.