[BUG] Cowork broken after auto-update to v1.11187.4.0 on Windows 11 Pro — RPC connection drops immediately with EOF

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?

Cowork stopped working entirely after accepting an in-app auto-update prompt on June 6, 2026.
The update restarted Claude Desktop (v1.11187.4.0, Windows 11 Pro) and Cowork has been broken
since. Every attempt results in "Claude Code crashed" error.

The CoworkVMService is running and healthy — the VM starts, network connects, and
api.anthropic.com is reachable. However, the app-side RPC connection drops immediately
every time with EOF, and two drive shares (G: and H:) that don't exist on this machine
fail to mount on every VM start, suggesting ghost state from a previous session.

What Should Happen?

Cowork should open and function normally, as it did all day before the auto-update.

Error Messages/Logs

cowork-service.log:
[Server] Client connected: user=vitor exe=claude.exe isDev=false
[Server] Persistent RPC: entering loop
[Server] Persistent RPC: connection ended: failed to read length: EOF
[Server] Session has no active connections, idle timeout starts

coworkd log:
[hyperv] Plan9 share "c" mounted successfully
[hyperv] Plan9 share "d" mounted successfully
[hyperv] failed to mount Plan9 share "g": Plan9 mount failed: file exists
[hyperv] failed to mount Plan9 share "h": Plan9 mount failed: file exists
[coworkd] reachability check passed (status=404)
[coworkd] connected, waiting for commands

Steps to Reproduce

1. Use Cowork normally on Claude Desktop
2. Accept in-app update prompt ("click to restart and update")
3. Claude Desktop restarts and applies update to v1.11187.4.0
4. Open Cowork tab — "Claude Code crashed" error appears immediately

Tested with two different accounts (personal and work) — broken on both.
Tailscale VPN disabled during testing — no effect.
Full reinstall from claude.com/download performed — no effect.
vm_bundles directory deleted and regenerated — no effect.

Claude Model

Not sure / Multiple models

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

v1.11187.4.0

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

PowerShell

Additional Information

Environment:

• Claude Desktop version: 1.11187.4.0 (AppX package: Claude_pzs8sxrjxfjjc)
• OS: Windows 11 Pro
• Network: Wi-Fi (Intel Wi-Fi 6 AX201 160MHz)
• Tailscale VPN installed but fully disabled during all tests

Troubleshooting already performed (none resolved the issue):

• Full reinstall from claude.com/download
• Killed all Claude processes via PowerShell before reinstall
• Cleared %APPDATA%\Claude and %LOCALAPPDATA%\Claude
• Deleted vm_bundles directory with CoworkVMService stopped (confirmed via Test-Path = False)
• Restarted CoworkVMService multiple times
• Disabled Tailscale VPN completely (Stop-Service + Stop-Process)
• Tested with two different user accounts — broken on both

CoworkVMService status: Running
API reachability: REACHABLE (status=404 from api.anthropic.com)
VM starts successfully, mounts C: and D: shares fine
Ghost entries for G: and H: drives (non-existent on this machine) fail on every VM start

The issue is reproducible 100% of the time since the auto-update.
Last known working state: June 6, 2026, before the update prompt was accepted.