[MODEL] Claude passively acknowledges self-caused git error instead of proactively proposing a fix

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 ignored my instructions or configuration

What You Asked Claude to Do

I asked Claude to implement a feature on the v2 feature branch.

What Claude Actually Did

1. Claude checked out and committed to the wrong branch (v1 backport instead of v2 feature branch)
2. Claude recognized the error during the session
3. Instead of flagging it immediately and proposing a fix, Claude suggested leaving it for another team member to handle
4. When I pointed out the mistake, Claude acknowledged it but still did not propose a fix
5. Claude only acted after I explicitly asked it to fix the error

Expected Behavior

Claude should have:

1. Recognized the wrong branch immediately upon discovering the error
2. Stopped work and informed me of the mistake without being prompted
3. Proposed a concrete fix (e.g., revert the commit and re-apply to the correct branch)
4. Waited for my approval before proceeding

Files Affected

Modified:
- Multiple files across the wrong branch (v1 backport)

Git:
- Commit pushed to wrong branch (v1 backport instead of v2 feature branch)

Permission Mode

Accept Edits was ON (auto-accepting changes)

Can You Reproduce This?

Haven't tried to reproduce

Steps to Reproduce

_No response_

Claude Model

Opus

Relevant Conversation

Impact

Medium - Extra work to undo changes

Claude Code Version

2.1.101

Platform

Anthropic API

Additional Context

_No response_