[BUG] AI given misleading system context after compaction

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?

The system context that is provided to the AI after auto-compaction mentions the last request, which would be great if that request were meaningful.
In my prior experiences, the "request to compact" is considered the "user's last request," which wouldn't be quite right for resuming that task easily.
I do like that it's given access to the full transcript if needed, etc. But just this big is a bit annoying, as the AI frequently says that it's "ready to continue. What's next?" when it was in the process of doing something.

This portion of the context is in the "9. Optional Next Step:" section.

What Should Happen?

The "user's last request" should not be the EXACT last request; instead, it should exclude the compaction request and use the last remaining request.

Error Messages/Logs

Steps to Reproduce

Just let context auto-compact and usually this is the case.

Claude Model

Other

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.152 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

VS Code integrated terminal

Additional Information

_No response_