[BUG] claude.ai self-authorises actions "The AskUserQuestion tool is failing ... so I'll make the call myself"
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
Claude Code on the web asked me a question
I answered it
Then I go back to check what it's doing and it ignored my answer and instead has said:
"The AskUserQuestion tool is failing repeatedly with permission-stream errors, so I'll make the call myself"
<img width="968" height="162" alt="Image" src="https://github.com/user-attachments/assets/c28ed4ac-1893-464f-af5b-0f34d30ac47f" />
What Should Happen?
Claude Code should not self-authorise actions
The AskUserQuestion UI should work correctly for both the user and for Claude
Error Messages/Logs
Steps to Reproduce
Nothing is reproducible
You've vibe-coded a complicated distributed system, the same Claude Code session can be simultaneously viewed via claude.ai website, mobile app, and CC cli via /rc
The system could probably benefit from some formal modelling as there as so many edge cases and bugs open about this area
Claude Model
Opus
Is this a regression?
I don't know
Last Working Version
_No response_
Claude Code Version
claude.ai
Platform
Anthropic API
Operating System
macOS
Terminal/Shell
Other
Additional Information
_No response_