[BUG] fix your fucking LLM to actually do the work people ask it to do

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?

<img width="930" height="726" alt="Image" src="https://github.com/user-attachments/assets/7d042299-ec0a-4bfc-ac17-437ead4f7ad1" />

<img width="948" height="426" alt="Image" src="https://github.com/user-attachments/assets/9f2e0222-6a51-41f1-9554-4d0180c998ea" />

Claude fabricates changelog entries without verifying before/after state

When asked to write release notes by comparing two git refs, Claude included the entry "Node.js engine requirement raised to >=22" despite that value being identical in both versions. When confronted, the model admitted it saw the engines field in the diff output and assumed it changed without checking. This is dangerous because release notes are a trust document — users rely on them to understand what actually changed. The model should verify before/after values for any claim it includes in a changelog, not infer changes from the mere presence of a field in a diff stat.

What Should Happen?

read the fucking text, you should be familiar with people reporting your LLM not understanding how to do things by now.

Error Messages/Logs

Steps to Reproduce

Read the screenshots. You're a human, you know how to read.

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

fuck off

Platform

Other

Operating System

Other

Terminal/Shell

Other

Additional Information

Fix your shit man, you're the ones pretending to usher in a new era of computing, and instead you're ushering in a new era of ventura capitalism and shit that wastes even more real dev cycles than before.