Claude Code Performance Degraded

Preflight Checklist

• [x] I have searched existing issues for similar behavior reports
• [x] This report does NOT contain sensitive information (API keys, passwords, etc.)

Type of Behavior Issue

Claude modified files I didn't ask it to modify

What You Asked Claude to Do

I asked Claude to check a file and explain why the ai wanted to update a schema. I had already checked what the result was prior because I noticed Claude wanted to update it without having any proof it was required. I noticed performance is severly degraded and logical abilities have dropped over the last 3 days considerably worse.

What Claude Actually Did

Claude lied about what the results were on the file and said there were 28 snippets that would be fixed by making the schema change that Claude was proposing. In fact there were none . Not a single one. I challenged at which point Claude admitted not checking and just having fabricated the data.

Expected Behavior

Its been losing the plot over the last few days. Not sure what happened

Files Affected

In this case i stopped claude before he could make the changes by catching him. He s been doing a number of crazy things all day long.

Permission Mode

Accept Edits was OFF (manual approval required)

Can You Reproduce This?

Haven't tried to reproduce

Steps to Reproduce

_No response_

Claude Model

Opus

Relevant Conversation

Opus 4.6

Impact

Medium - Extra work to undo changes

Claude Code Version

4.6

Platform

Anthropic API

Additional Context

I am just flagging this as a model degradation .. I have been using it heavily but over the last 3 days its become very laborious to get it to work efficiently. I had to put a bunch of blocks in place to avoid damage. I put a cron block and a web hook to stop ovverwriting work