[BUG] Switching branches despite memory to NOT

Resolved 💬 4 comments Opened Mar 12, 2026 by samc-heretto Closed May 1, 2026

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?

I was working on a problem, and for the third time claude decided to switch branches to resolve a test issue, but has done int he past for a compile issue, despite an explicit memory to NEVER switch branches. I lost a lot of work, so that cost me tokens, then I just used a LOT of tokens to try to recover it. I get it that things happen, but at very least, when claude makes a mistake like that I shouldn't have to use tokens to recover it, which it's failed to yet.

What Should Happen?

I had told claude never to switch branches for this, in fact right it had and dropped the files, it said "Oh you said never to do this" and switch another branch wrong one, and made an ever bigger mess, in the middle the context go reset and the transcripts are lost - claude should ALWAYS keep backups, I don't care about the disk space, but losing work, i would glad pay extra for that level of security.

Error Messages/Logs

It reset the context, clearing the screen (why it does that I don't know) so I can't get them for you

Steps to Reproduce

Not sure I have good steps

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

⏺ Bash(claude --version 2>&1) ⎿  2.1.74 (Claude Code) ⏺ Version 2.1.74.

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

iTerm2

Additional Information

_No response_

View original on GitHub ↗

This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗