[BUG] Feedback harness collides with Claude's numbered responses.

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?

Your feedback harness collides with Claude's numbered responses. I intend to respond to the response labeled "2" and instead answer "how is claude doing" before I even realize the harness intrusion was there. You are getting wildly polluted data by intruding mid-conversation and using the same pathway to get that feedback. If you want correct appraisals, you should probably use a different route or make it so that the numbered responses you are reading don't conflict with the active conversation.

What Should Happen?

Feedback intrusion in the harness should come in via a different path or numbers shouldn't conflict with Claude's previous response.

Error Messages/Logs

Steps to Reproduce

1. get into a conversation
2. start discussing code
3. Claude responds with a numbered list
4. feedback intrusion
5. I intend to enumerate a response "1. This is not accurate due to...." and instead my keystrokes get lost and instead I vote "Claude is doing bad"
6. You get bad data.

Claude Model

Not sure / Multiple models

Is this a regression?

No, this never worked

Last Working Version

_No response_

Claude Code Version

v2.1.198

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

IntelliJ IDEA terminal

Additional Information

I'm just trying to let you know that your feedback intrusion is giving you more bad data than you realize.