Hey, thanks for the report. This is a known issue with loop detection. Sometimes the detector logic is too aggressive and stops the agent even when it’s not actually stuck in a loop.
In your case, the agent got stuck trying to re-add init to the mock on line 160, which looks like a false positive from the detector.
The team is working on improving how it detects real loops. A similar case was discussed here: Too aggressive loop detection
Workaround:
Instead of clicking Resume, try typing “continue” in the chat. This often resets the internal loop counter and lets the agent keep going.
Request ID:
To pass this case to the team, we need the Request ID, but I understand you’re using a work account with Privacy Mode. If you run into this again and can temporarily disable Privacy Mode, the Request ID will help the engineers review the specific case and improve the detector.