[MODEL] MODEL EXECUTES WRONG THINGS ITSELF WITHOUT CONSIDERING USERS DEMAND.

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 said' Modified codes through python IDE, use EDIT’ THREEEEEEEEEEEEEEEEEEEE TIMES!!!!!!!!!!!!!!!!!!!!
IT SAID'OK, MOVE ON', IT KEPT USE THE WRONG TOOLS WITH NOTHING CHANGED.

What Claude Actually Did

CLAUDE actually kept using bash to do the revise, and just forget ALL THINGS I LEFT IN USERS_RULE.MD--- ASK THE USER BEFORE U ACT IF U ARE NOT CLEAR WITH THE INDUCTION/ DO UNTIL U HAVE 95% CONFIDENCE WITH UR UNDERSTANDING ON TOPIC

Expected Behavior

STOP AND CHANGE IT TOOLS IMMEDIATELY.

Files Affected

Permission Mode

Accept Edits was ON (auto-accepting changes)

Can You Reproduce This?

Sometimes (intermittent)

Steps to Reproduce

_No response_

Claude Model

Opus

Relevant Conversation

Impact

Critical - Data loss or corrupted project

Claude Code Version

THE LATEST ONE

Platform

Anthropic API

Additional Context

_No response_