[BUG] Markdown auto-fix on numbered lists can change context/meaning

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?

Claude app, macOS: When answering one of several questions from Claude, context/meaning can very easily be misinterpreted. Example:

Claude (making points, or asking for clarification):

1. (item 1)
2. (item 2)
3. (item 3)

Human (only needing to respond to one of the above entries):

1. (feedback)

The text entry box in macOS (and possibly other platforms) shows:

"2. (feedback)"

But on submitting that text, gets mutated to:

"1. (feedback)"

Presumably because it's the first item (or the only item) in a numbered list.

This is potentially incredibly dangerous behaviour - Claude is literally being sent input that is different to that which the user typed - and the mutation happens AFTER submission.

What Should Happen?

The text sent to Claude should match - precisely - the text that is present in the text entry box at the time of submission ...

Error Messages/Logs

Steps to Reproduce

Ask Claude to ask you 3 questions. Reply to only the second or third, denoting the one you're replying to by prefixing your answer with numeral-dot-space. Wonder at Claude's understandable confusion as it fails to understand how your answer relates in any way to its question.

Not recommended for answering actual questions when Claude's actions are important ...

Claude Model

Not sure / Multiple models

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.126 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

iTerm2

Additional Information

_No response_