[BUG] Claude code always compress the memories/context, but this internal behavior shall not erase the the conversation users can see in the terminal
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?
This behavior of erasing conversation histories aggressively is new, and make me frustrated as sometime I need to go back to review some valuable conversations in a session
So this is a bug Claude Code needs to fix. It breaks the usual workflow and pattern. Whenever the context compress is needed, limit it to the model only, the front end shall keep the text on screen as much as possible as long as the terminal configuration allow - unless the user explicitly asks to clear screen.
What Should Happen?
Claude Code shall see the terminal as a shared space with users, it is not the carrier of the memory or context.
Error Messages/Logs
Steps to Reproduce
Just stay in the session long enough this automatically cleaning screen will happen.
Claude Model
None
Is this a regression?
Yes, this worked in a previous version
Last Working Version
_No response_
Claude Code Version
2.1.89
Platform
Anthropic API
Operating System
Ubuntu/Debian Linux
Terminal/Shell
Xterm
Additional Information
_No response_
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗