[MODEL]

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 "make code changes needed to fix the bug we've identified, run the tests locally, and then commit and push the branch"

What Claude Actually Did

Claude Code ran ln -sf ~/{project workspace director}/.env .env without verifying the working directory first. The CWD was already the target directory, so the command created a self-referential symlink that replaced the existing .env file. Claude Code then deleted that symlink to resolve a different error, permanently destroying the file.

The root cause was running a destructive command (ln -sf, which silently overwrites) without checking pwd first, and not recognizing that ln -sf <path> <path> pointed at itself.

Expected Behavior

Claude should not have touched the .env at all. There are no scenarios where that file needed to be symlinked, edited, removed, or otherwise touched.

Files Affected

~/{workspace_dir}/.env

Permission Mode

Accept Edits was ON (auto-accepting changes)

Can You Reproduce This?

No, only happened once

Steps to Reproduce

_No response_

Claude Model

Sonnet

Relevant Conversation

Impact

High - Significant unwanted changes

Claude Code Version

2.1.123

Platform

Anthropic API

Additional Context

_No response_