Cowork Windows - VM gets wrong DNS server, API unreachable

Resolved 💬 3 comments Opened Feb 11, 2026 by betsyalter Closed Feb 15, 2026

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 workspace boots successfully but the VM cannot reach the Claude API. The VM is assigned the wrong DNS server (192.168.1.254) instead of the correct host DNS (192.168.1.1). This causes api.anthropic.com to be unresolvable from inside the VM. Every startup attempt follows the same pattern: VM boots → SDK installs → API reachability goes from PROBABLY_UNREACHABLE → UNREACHABLE.
This is the same root cause as #24962 (wrong DNS configuration in HNS endpoint).

What Should Happen?

The VM should inherit the correct DNS server from the host network (192.168.1.1) and successfully reach api.anthropic.com.

Error Messages/Logs

Failed to start Claude's workspace - VM service not running. The service failed to start.

From cowork_vm_node.log:
2026-02-11 12:04:34 [info] [VM:start] Configuring Windows VM service...
2026-02-11 12:04:35 [info] [VM:start] Windows VM service configured
2026-02-11 12:04:46 [info] [VM] Network status: NOT_CONNECTED
2026-02-11 12:04:47 [info] [VM] Network status: CONNECTED
2026-02-11 12:04:52 [info] [VM] API reachability: PROBABLY_UNREACHABLE
2026-02-11 12:04:53 [info] [VM:start] Startup complete, total time: 18728ms
2026-02-11 12:05:16 [info] [VM] API reachability: UNREACHABLE
2026-02-11 12:05:16 [warn] [VM:network] API is unreachable

DNS mismatch evidence (from PowerShell):
# VM endpoint - WRONG DNS
Name           : cowork-vm-eth0
IPAddress      : 172.16.0.171
GatewayAddress : 172.16.0.1
DNSServerList  : 192.168.1.254

# Host adapter - CORRECT DNS
InterfaceAlias : Ethernet 2
ServerAddresses: 192.168.1.1

Test-NetConnection api.anthropic.com -Port 443 → TcpTestSucceeded: True

Steps to Reproduce

  1. Install Claude Desktop on Windows 11 Pro for Workstations (Build 26200.7840)
  2. Open Claude Desktop
  3. Switch to Cowork tab
  4. Attempt to start any task
  5. Wait ~60 seconds
  6. Observe "VM service not running" error
  7. Run Get-HnsEndpoint in admin PowerShell to confirm DNS mismatch

Claude Model

None

Is this a regression?

No, this never worked

Last Working Version

_No response_

Claude Code Version

Claude Desktop 1.1.2685.0 (Windows MSIX)

Platform

Other

Operating System

Windows

Terminal/Shell

PowerShell

Additional Information

Troubleshooting attempted (all failed to resolve):

Reinstalled Claude Desktop multiple times (complete uninstall + fresh install)
Deleted vm_bundles from %AppData%\Claude\ and %LocalAppData%\Claude\
Deleted and let Claude recreate the HNS cowork-vm-nat network
Restarted Hyper-V services (vmcompute, vmms)
Verified all Windows virtualization features enabled (Virtual Machine Platform, Windows Hypervisor Platform, Hyper-V)
Confirmed vmcompute and vmms services running
Confirmed host can reach api.anthropic.com on port 443

Operating System: Windows 11 Pro for Workstations, Build 26200.7840
Terminal/Shell: Windows PowerShell 5.1 / PowerShell 7

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗