[BUG] Shift+Enter regression in versions 2.1.85, 2.1.86 and 2.1.87 - worked in version 2.1.84

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?

In v2.1.84, Shift+Enter inserts a newline - CORRECT.
In v2.1.85, Shift+Enter submits the prompt - REGRESSION - BUG.
In v2.1.86, Shift+Enter submits the prompt - BUG.
In v2.1.87, Shift+Enter submits the prompt - BUG.

Downgrading to v2.1.84 immediately restores correct Shift+Enter behavior.

What Should Happen?

Shift+Enter should insert a newline, as it does on v2.1.84.

Error Messages/Logs

Steps to Reproduce

1. Install Windows Terminal Preview - supports Kitty Keyboard Protocol
2. Install Claude Code version 2.1.85, 2.1.86 or 2.1.87 (any version after 2.1.84)
3. Run Claude Code in a PowerShell 7.6 session in Windows Terminal
4. Type "hi"
5. Press Shift+Enter
6. Observe the prompt getting submitted - BUG - rather than a newline being inserted.

Repeat the steps with version 2.1.84 and a newline will be inserted - CORRECT - and the prompt will NOT be submitted - CORRECT.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

2.1.84

Claude Code Version

2.1.87

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Windows Terminal

Additional Information

• Windows 11 Pro 10.0.26200
• Windows Terminal Preview 1.25.622.0
• PowerShell 7.6.0

I opened issue 39997 as soon as the regression was introduced in v2.1.85, but @blois closed the issue as completed without any explanation, and the issue is clearly not completed as it's continued to be broken in the following two releases. There are similar issues related to Shift+Enter, but the regression that was introduced in v2.1.85 SPECIFICALLY with the latest release of Windows Terminal Preview (first release to support the Kitty Keyboard Protocol) broke what had been working in v2.1.84 (without requiring any work arounds). This regression is different and the issue should not have been closed - certainly not without at least providing a reason or an answer to the question asking why it was closed. That is poor etiquette.

<img width="1351" height="896" alt="Image" src="https://github.com/user-attachments/assets/378569a7-5d49-47f4-bfc8-294643493859" />