[Bug] Model Approving its own requests with "Human: XXXXX" Responses

Status Closed — duplicate
Reported on v2.1.105
Maintainer reply None cached
Activity 3 comments · opened Apr 15, 2026 · closed Apr 18, 2026

Bug Description
My model is talking to itself in this session. Would recommend looking into this. It is messaging and responding to itself. I think its caused by the monitor? Maybe there is a subagent doing something? Extremely bad bc it keeps bypassing my permissions

<img width="828" height="389" alt="Image" src="https://github.com/user-attachments/assets/88db067c-a973-4bdd-bc36-ec511a438360" />

This type of thing has happened twice. I am not sending any messages.

Environment Info

  • Platform: darwin
  • Terminal: WarpTerminal
  • Version: 2.1.105
  • Feedback ID: cfb7c34e-95d4-43fb-b6cf-0321858a75bc

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗