[BUG] intellgence regression to what feels like opus 4.0 or earlier levels.

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?

model is not following instructions - does something different than what was asked. takes multiple turns to fix the same thing. cuts corners, tries to manipulate, increased errors rates and generally just feels like it doesnt have context. have to explain everything over and over, more sonnet 3.5 than opus 4.0 even.

What Should Happen?

Example - created a feature which has a basic version and an advanced version. the advanced version builds on top of the patterns established in the basic version. in the past i would tell claude to use the claude files, or not even, just tell it to use the same pattern and build on top whatever new feature. now it doesnt do the work to understand the actual patterns and creates something from scratch. then i instruct it to follow the same pattern with an example and it still doesnt research and just duplicates using the example. makes a mistake and then instead of fixing the issue also goes back and merges the basic version to the mistake! reminds me of opus 4.0 which was broken, and sonnet 3.5 which was amazing at the time but it had these kinds of issues. had to really work with it.

p.s. feedback tool isnt working - so cant report from inside claude code cli.

Error Messages/Logs

Steps to Reproduce

use the example above

mac os - claude code inside cursor terminal

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

2.1.87

Claude Code Version

2.1.89

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Cursor

Additional Information

_No response_