[BUG] Claude sending commands/messages to subagents inside the terminal and subagent not receiving it

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?

When claude invokes a subagent, often the subagent will stop and ask for confirmation on something. Claude will often reply to the agent in the terminal conversation window with the user, rather than sending the command to the subagent.

What Should Happen?

Claude should send it's commands diretly to the subagent, so the subagent can continue it's work.

Error Messages/Logs

Steps to Reproduce

Launch a subagent and wait for it to ask for confirmation on what to do.

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

v2.0.33

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

WSL (Windows Subsystem for Linux)

Additional Information

<img width="1097" height="163" alt="Image" src="https://github.com/user-attachments/assets/d0dfb423-6046-4988-94da-79840e807c42" />

<img width="1745" height="370" alt="Image" src="https://github.com/user-attachments/assets/d8b26aca-4bfe-41b8-99db-c3a019a61bfd" />