Repository navigation
[BUG] [Cowork] Windows: VM starts successfully but crashes within 5 minutes, service unrecoverable (Win 11 Pro, Hyper-V) #25206
Description
Preflight Checklist
- I have searched existing issues and this hasn't been reported yet
- This is a single bug report (please file separate reports for different bugs)
- I am using the latest version of Claude Code
What's Wrong?
Preflight Checklist
- I have searched existing issues and this hasn't been reported yet
- This is a single bug report
- I am using the latest version of Claude Code
What's Wrong?
Cowork workspace setup begins but consistently fails with: "VM service not running. The service failed to start."
The VM briefly starts successfully (confirmed via Hyper-V event logs) but crashes within ~5 minutes. All subsequent attempts fail with the same error. This appears related to the broader pattern of Windows Cowork VM failures reported in #24974, #24962, #25024, #24945, and #24918.
Environment
- Claude Desktop: v1.1.2685
- Windows 11 Pro (build 26220)
- Lenovo ThinkPad X1 Nano Gen 1 (20XYS03200)
- 11th Gen Intel i7-1165G7, 16 GB RAM
- Hyper-V enabled, Intel VT-x enabled in BIOS
- Hyper-V services (vmms, vmcompute) running
Diagnostic Steps Taken
- Enabled Hyper-V via
Enable-WindowsOptionalFeature -Online -FeatureName Microsoft-Hyper-V -All - Enabled Intel Virtualization Technology (VT-x) in BIOS (Security > Virtualization)
- Confirmed Hyper-V services running: vmms (Automatic), vmcompute (Manual)
- Confirmed
VirtualizationFirmwareEnabled = TrueandVMMonitorModeExtensions = True - Deleted claude-code-vm and vm_bundles directories to force clean rebuild — same failure
- Multiple Claude app restarts and full system reboots — no change
Hyper-V Event Log Evidence
The VM ('cowork-vm', ID 960E110C-FC59-5B4B-843D-68737D617893) started successfully at 13:55:21, connected to virtual network, then was turned off by 14:00:24. Subsequent attempts fail to restart the service.
13:55:20 - 'cowork-vm' properties were successfully initialized
13:55:21 - 'cowork-vm' connected to virtual network
13:55:21 - 'cowork-vm' started successfully
14:00:24 - 'cowork-vm' was turned off
Sentry Error IDs from main.log
504336fc02cb420495984bc29c96127dfdb50d1f08454b29a3cbeb235b3f0d15
Both errors: "VM service not running. The service failed to start."
Additional Log Notes
- CSP violation errors blocking connections to
https://a-api.anthropic.com/v1/m— consistent with the CSP bug reported in [BUG] Cowork Windows - CSP blocks a-api.anthropic.com, VM starts but API unreachable (Win 11 25H2) #25024 (fixed on macOS but not ported to Windows) - Repeated
MaxListenersExceededWarningfor AutoUpdater state store - Bundle version:
fc96f5bc6a7aeee6e651437eba3c76d0fe9a16f3
Related Issues
- [BUG] [Cowork] Windows: sessiondata.vhdx not created + network diagnostics uses macOS commands #24974 —
sessiondata.vhdxnot created + macOS diagnostics on Windows - Cowork VM fails on Windows 11 Home: wrong DNS config, missing sessiondata.vhdx, no NAT #24962 — Missing sessiondata.vhdx, no NAT, macOS commands on Windows
- [BUG] Cowork Windows - CSP blocks a-api.anthropic.com, VM starts but API unreachable (Win 11 25H2) #25024 — CSP blocks
a-api.anthropic.com, API unreachable from VM - [BUG] Cowork VM boots successfully but API is unreachable on Windows #24945 — VM boots but API unreachable on Windows
- [BUG] Cowork Windows - Cannot connect to Claude API from workspace (Windows 11 Home) #24918 — Cannot connect to Claude API from workspace (Windows 11)
Expected Behaviour
Cowork VM should start and remain running, with the workspace available for task execution.
What Actually Happens
VM starts, crashes within minutes, and cannot be restarted without manual cleanup of app data directories — after which the rebuilt VM crashes identically.
What Should Happen?
After Hyper-V and VT-x are enabled, the Cowork VM should start, remain running for the duration of the task, and recover gracefully if interrupted. Specifically:
- The VM should persist beyond initial boot without crashing unprompted
- If the VM service fails, the app should be able to restart it without requiring manual deletion of
claude-code-vmandvm_bundlesdirectories - The Cowork tab should surface actionable error information rather than the generic "VM service not running" message, particularly any guest OS or coworkd logs from inside the VM that would explain why it stopped
Error Messages/Logs
In Cowork:
Failed to start Claude's workspace
VM service not running. The service failed to start.
Restarting Claude or your computer sometimes resolves this. If it persists, you can reinstall the workspace or share your debug logs to help us improve.Steps to Reproduce
- Install Claude Desktop v1.1.2685 on Windows 11 Pro
- Enable Hyper-V and Intel VT-x (required — neither was enabled by default)
- Open Cowork and submit a task
- Workspace setup begins, VM starts successfully
- VM crashes within ~5 minutes
- All subsequent attempts fail with "VM service not running. The service failed to start."
- Deleting claude-code-vm and vm_bundles directories triggers a clean rebuild, but the rebuilt VM crashes identically
Claude Model
Opus
Is this a regression?
No, this never worked
Last Working Version
No response
Claude Code Version
Claude Desktop: v1.1.2685
Platform
Anthropic API
Operating System
Windows
Terminal/Shell
PowerShell
Additional Information
No response
Activity
github-actions commentedon Feb 12, 2026
Found 3 possible duplicate issues:
This issue will be automatically closed as a duplicate in 3 days.
🤖 Generated with Claude Code
thisisJohnAU commentedon Feb 12, 2026
This is not a duplicate of the flagged issues. Those three reports (#25024, #25155, #24945) describe a scenario where the VM boots and stays running but the API is unreachable from inside the VM, a networking/CSP problem.
My issue is different - the VM itself crashes within ~5 minutes of starting (confirmed via Hyper-V event logs showing the VM was turned off at 14:00:24 after starting at 13:55:21). The failure is at the VM lifecycle level, not the network layer. After the crash, the VM service cannot be restarted at all, even after deleting and rebuilding the VM bundle.
The CSP violation errors in my logs may be a contributing factor (and may overlap with #25024), but the core problem is distinct: VM process termination, not API unreachability.
JannetjeIQ commentedon Feb 15, 2026
Also experiencing this on Windows 11 Pro Insider Preview Build 26200.5710.
Same symptoms: VM service fails to start, missing sessiondata.vhdx and smol-bin.vhdx from bundle directory, no VM appears in Hyper-V Manager.
Full system specs: Intel i9-9900, 64GB RAM, Claude Desktop 1.1.3189.0
Tried all troubleshooting steps (disabled disconnected adapters, ran as admin, multiple reinstalls) - none resolved the issue. Debug logs shared via Claude Desktop.
FrankvD-Prog02 commentedon Mar 1, 2026
**+1 — Still experiencing this on the latest version.
Environment:
Plan9 mount failed: bad addressandsignature verification failed: client executable is not signedin logs.cowork_vm_node.log:github-actions commentedon Mar 29, 2026
Closing for now — inactive for too long. Please open a new issue if this is still relevant.
github-actions commentedon Apr 6, 2026
This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.