claude-code - 💡(How to fix) Fix [Bug] Model Approving its own requests with "Human: XXXXX" Responses [1 comments, 2 participants]

TL;DR

The issue may be resolved by investigating and potentially disabling any subagents that could be causing the model to message itself, bypassing permissions.

Guidance

• Investigate the role of the monitor and subagents in the model's behavior to determine if they are contributing to the self-messaging issue.
• Review the model's configuration and permissions to ensure that they are correctly set up and not being bypassed by the subagents.
• Check for any similar issues in the model's history, as this has happened twice before, to identify potential patterns or triggers.
• Consider testing the model in a controlled environment to isolate the cause of the issue and verify that it is not related to the platform, terminal, or version.

Example

No code snippet is provided as the issue does not contain sufficient technical details to generate a relevant example.

Notes

The issue lacks detailed technical information, making it challenging to provide a definitive solution. The suggestions provided are based on the limited information available and may need to be adapted or refined as more details become available.

Recommendation

Apply workaround: Investigate and potentially disable subagents, as this is the most likely cause of the issue based on the provided information. This approach allows for a targeted investigation and potential resolution without requiring significant changes to the model or its environment.