claude-code - 💡(How to fix) Fix VM connection timeout after 60 seconds

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?

After doing all these:-

Hi! I am an AI-powered chatbot for Anthropic. Your use of this chatbot and chat data will be monitored, retained, and used by Anthropic, and our third-party chatbot provider (Intercom), to provide you with better customer support, to improve our services, and for other purposes consistent with our privacy policy (link). Please indicate whether you agree to these conditions of use. Fin AI Agent • AI Agent • 1h Accept Sukhen • 1h Conversation ID for your records: 215474388561877 Let us know what your inquiry is about. Fin AI Agent • AI Agent • 1h Technical Issues/Errors Sukhen • 1h Conversation ID for your records: 215474388561877 Go ahead and provide the details of your inquiry. Fin AI Agent • AI Agent • 1h Error: "VM connection timeout after 60 seconds" Device: ASUS Vivobook S16, ARM64, Snapdragon X, Windows 11 Steps already tried: uninstall, registry clean, folder deletion, fresh reinstall as admin Attach the log file that opened in Explorer when the error first appeared Sukhen • 1h The "VM connection timeout after 60 seconds" error you're experiencing isn't related to any current system outages. This timeout issue appears to be specific to MCP (Model Context Protocol) server connections in Claude Code. For your ARM64 Snapdragon X system, try these additional steps: Network connectivity check: Run this in PowerShell to verify you can reach our servers: curl -sI https://downloads.claude.ai/claude-code-releases/latest

You should see an HTTP/2 200 response. TLS configuration: Enable TLS 1.2 in PowerShell before reinstalling: [Net.ServicePointManager]::SecurityProtocol = [Net.SecurityProtocolType]::Tls12

Check for network restrictions: Since custom connectors connect through our cloud infrastructure rather than locally, corporate firewalls or network restrictions could be blocking the connection. If you're on a corporate or restricted network, this might be the cause. Force close any remaining processes: Use Task Manager (Ctrl+Shift+Esc) to ensure no Claude processes are still running before reinstalling. The 60-second timeout is the default for MCP server connections, so the error suggests a connection issue rather than a software problem. Fin AI Agent • AI Agent • 1h Is that what you were looking for? Fin AI Agent • 1h We noticed you might have stepped away. We'll close this conversation for now, but if you need additional assistance you can always follow up here. Fin AI Agent • 59m

Thanks for sharing the screenshot. The "bash failed on resume, create, and re-resume" errors with 30-second timeouts point to the workspace VM struggling to initialize properly on your ARM64 system. Since you've already done a clean reinstall, here are some additional steps specific to workspace VM issues: Check Windows Event Viewer logs: Open Event Viewer → Windows Logs → Application and look for Claude-related errors around the time of the timeout. This can reveal if Windows is blocking VM processes. Verify no conflicting Claude installations: Run this in PowerShell to check for multiple installations: Test-Path "$env:USERPROFILE.local\bin\claude.exe"

If you find extras from npm, Homebrew-like tools, or old installs, remove them as they can interfere with workspace initialization. Network/proxy interference: The 30-second bash timeouts suggest network issues during VM setup. If you're on a corporate network or using a VPN, try temporarily disabling them. The workspace needs to reach our infrastructure without proxy interference. File a bug report: Since this is persisting after reinstall on ARM64 Snapdragon X (which is newer hardware), consider filing a bug on GitHub Issues with your screenshot, Windows version, and Event Viewer logs. Include that you're on ARM64 Snapdragon X specifically. The 60-second VM timeout is hardcoded, so if the workspace can't initialize within that window, it will fail. Fin AI Agent • AI Agent • 33m Did that answer your question? Fin AI Agent • 33m

Good — that confirms the native installation exists at the expected location. Since Test-Path returned True, this is actually the correct installation path for the native installer. Now check for additional conflicting installations that might be interfering: Check for legacy npm install: Test-Path "$env:USERPROFILE.claude\local"

Check for multiple claude.exe files: where.exe claude

If where.exe claude shows more than one path, or if the legacy .claude\local folder exists, you'll need to remove the extras while keeping the native install at %USERPROFILE%.local\bin\claude.exe. To remove the legacy npm install if found: Remove-Item -Recurse -Force "$env:USERPROFILE.claude\local"

After cleaning up any conflicting installations, restart your computer and try launching Claude Code again. The VM timeout might be caused by multiple installations trying to initialize workspaces simultaneously. Fin AI Agent • AI Agent • 31m Was this helpful? Fin AI Agent • 31m

Good — no conflicting installations found. Since you have only the native install and you've already done a clean reinstall, the VM timeout is likely network-related. Next steps to try: Run where.exe claude to confirm there's truly only one executable path. If it shows multiple, remove the extras. Check your network setup: If you're on a corporate network or VPN, try disabling them temporarily. The workspace VM connects through our cloud infrastructure and can be blocked by firewalls or proxies. Set your proxy if needed: $env:HTTPS_PROXY="your-proxy-address" before launching Claude Code. Increase timeout tolerance: Since ARM64 Snapdragon X is newer hardware, the VM might need more startup time. Try setting: $env:MCP_CONNECT_TIMEOUT_MS=10000

This doubles the connection wait time from 5 to 10 seconds. If the issue persists after these steps, it's worth reporting on GitHub Issues with your system specs and Event Viewer logs, as this could be ARM64-specific. Fin AI Agent • AI Agent • 29m Did this help resolve the timeout issue? Fin AI Agent • 29m PS C:\Windows\system32> where.exe claude C:\Users\sukhe.local\bin\claude.exe PS C:\Windows\system32> ​ Sukhen • 28m Perfect — only one Claude installation confirmed. Since you've ruled out conflicting installations and the VM timeout persists, this points to a network or ARM64-specific issue. Try these final steps: Set the MCP connection timeout environment variable: $env:MCP_CONNECT_TIMEOUT_MS=120000

This extends the 60-second VM connection timeout to 2 minutes, which may help on ARM64 hardware. Check Windows Event Viewer for Claude-related errors: Open Event Viewer → Windows Logs → Application and look for errors around the time of the timeout. This can reveal if Windows is blocking VM processes. Verify network access: If you're on a corporate network or VPN, try disabling them temporarily. The workspace VM connects through our cloud infrastructure and can be blocked by firewalls or proxies. If the issue continues after these steps, consider reporting it on GitHub Issues with your system specs (ARM64 Snapdragon X, Windows 11), the screenshot you shared, and any Event Viewer logs — this could be ARM64-specific behavior that needs investigation. Fin AI Agent • AI Agent • 28m Did that help with the timeout issue? Fin AI Agent • 28m We noticed you might have stepped away. We'll close this conversation for now, but if you need additional assistance you can always follow up here. Fin AI Agent • 18m No., it is showing again Sukhen • 8m New Since the VM timeout persists on your ARM64 Snapdragon X system after all standard troubleshooting, this appears to be a hardware-specific issue that needs developer attention. Report this on GitHub: File a bug at https://github.com/anthropics/claude-code/issues with: Your screenshot showing the bash timeout errors System specs: ASUS Vivobook S16, ARM64 Snapdragon X, Windows 11 Steps you've tried (clean reinstall, registry clean, environment variables) Any Windows Event Viewer logs from around the timeout The 60-second VM connection timeout is hardcoded in the MCP SDK, and ARM64 Snapdragon X is newer hardware that may need additional startup time or compatibility fixes.

IT IS STILL THERE <img width="1197" height="807" alt="Image" src="https://github.com/user-attachments/assets/3c84910e-163f-4e2b-a51e-bc4c347ab2f1" />

What Should Happen?

Cowork is not able to execute shell, bash scripts.

Error Messages/Logs

Hi! I am an AI-powered chatbot for Anthropic.
Your use of this chatbot and chat data will be monitored, retained, and used by Anthropic, and our third-party chatbot provider (Intercom), to provide you with better customer support, to improve our services, and for other purposes consistent with our privacy policy (link).
Please indicate whether you agree to these conditions of use.
Fin AI Agent • AI Agent • 1h
Accept
Sukhen • 1h
Conversation ID for your records: 215474388561877
Let us know what your inquiry is about.
Fin AI Agent • AI Agent • 1h
Technical Issues/Errors
Sukhen • 1h
Conversation ID for your records: 215474388561877
Go ahead and provide the details of your inquiry.
Fin AI Agent • AI Agent • 1h
Error: "VM connection timeout after 60 seconds"
Device: ASUS Vivobook S16, ARM64, Snapdragon X, Windows 11
Steps already tried: uninstall, registry clean, folder deletion, fresh reinstall as admin
Attach the log file that opened in Explorer when the error first appeared
Sukhen • 1h
The "VM connection timeout after 60 seconds" error you're experiencing isn't related to any current system outages. This timeout issue appears to be specific to MCP (Model Context Protocol) server connections in Claude Code.
For your ARM64 Snapdragon X system, try these additional steps:
Network connectivity check:
Run this in PowerShell to verify you can reach our servers:
curl -sI https://downloads.claude.ai/claude-code-releases/latest

You should see an HTTP/2 200 response.
TLS configuration:
Enable TLS 1.2 in PowerShell before reinstalling:
[Net.ServicePointManager]::SecurityProtocol = [Net.SecurityProtocolType]::Tls12

Check for network restrictions:
Since custom connectors connect through our cloud infrastructure rather than locally, corporate firewalls or network restrictions could be blocking the connection. If you're on a corporate or restricted network, this might be the cause.
Force close any remaining processes:
Use Task Manager (Ctrl+Shift+Esc) to ensure no Claude processes are still running before reinstalling.
The 60-second timeout is the default for MCP server connections, so the error suggests a connection issue rather than a software problem.
Fin AI Agent • AI Agent • 1h
Is that what you were looking for?
Fin AI Agent • 1h
We noticed you might have stepped away. We'll close this conversation for now, but if you need additional assistance you can always follow up here.
Fin AI Agent • 59m

Thanks for sharing the screenshot. The "bash failed on resume, create, and re-resume" errors with 30-second timeouts point to the workspace VM struggling to initialize properly on your ARM64 system.
Since you've already done a clean reinstall, here are some additional steps specific to workspace VM issues:
Check Windows Event Viewer logs:
Open Event Viewer → Windows Logs → Application and look for Claude-related errors around the time of the timeout. This can reveal if Windows is blocking VM processes.
Verify no conflicting Claude installations:
Run this in PowerShell to check for multiple installations:
Test-Path "$env:USERPROFILE\.local\bin\claude.exe"

If you find extras from npm, Homebrew-like tools, or old installs, remove them as they can interfere with workspace initialization.
Network/proxy interference:
The 30-second bash timeouts suggest network issues during VM setup. If you're on a corporate network or using a VPN, try temporarily disabling them. The workspace needs to reach our infrastructure without proxy interference.
File a bug report:
Since this is persisting after reinstall on ARM64 Snapdragon X (which is newer hardware), consider filing a bug on GitHub Issues with your screenshot, Windows version, and Event Viewer logs. Include that you're on ARM64 Snapdragon X specifically.
The 60-second VM timeout is hardcoded, so if the workspace can't initialize within that window, it will fail.
Fin AI Agent • AI Agent • 33m
Did that answer your question?
Fin AI Agent • 33m

Good — that confirms the native installation exists at the expected location. Since Test-Path returned True, this is actually the correct installation path for the native installer.
Now check for additional conflicting installations that might be interfering:
Check for legacy npm install:
Test-Path "$env:USERPROFILE\.claude\local"

Check for multiple claude.exe files:
where.exe claude

If where.exe claude shows more than one path, or if the legacy .claude\local folder exists, you'll need to remove the extras while keeping the native install at %USERPROFILE%\.local\bin\claude.exe.
To remove the legacy npm install if found:
Remove-Item -Recurse -Force "$env:USERPROFILE\.claude\local"

After cleaning up any conflicting installations, restart your computer and try launching Claude Code again. The VM timeout might be caused by multiple installations trying to initialize workspaces simultaneously.
Fin AI Agent • AI Agent • 31m
Was this helpful?
Fin AI Agent • 31m

Good — no conflicting installations found. Since you have only the native install and you've already done a clean reinstall, the VM timeout is likely network-related.
Next steps to try:
Run where.exe claude to confirm there's truly only one executable path. If it shows multiple, remove the extras.
Check your network setup:
If you're on a corporate network or VPN, try disabling them temporarily. The workspace VM connects through our cloud infrastructure and can be blocked by firewalls or proxies.
Set your proxy if needed: $env:HTTPS_PROXY="your-proxy-address" before launching Claude Code.
Increase timeout tolerance:
Since ARM64 Snapdragon X is newer hardware, the VM might need more startup time. Try setting:
$env:MCP_CONNECT_TIMEOUT_MS=10000

This doubles the connection wait time from 5 to 10 seconds.
If the issue persists after these steps, it's worth reporting on GitHub Issues with your system specs and Event Viewer logs, as this could be ARM64-specific.
Fin AI Agent • AI Agent • 29m
Did this help resolve the timeout issue?
Fin AI Agent • 29m
PS C:\Windows\system32> where.exe claude
C:\Users\sukhe\.local\bin\claude.exe
PS C:\Windows\system32>
​
Sukhen • 28m
Perfect — only one Claude installation confirmed. Since you've ruled out conflicting installations and the VM timeout persists, this points to a network or ARM64-specific issue.
Try these final steps:
Set the MCP connection timeout environment variable:
$env:MCP_CONNECT_TIMEOUT_MS=120000

This extends the 60-second VM connection timeout to 2 minutes, which may help on ARM64 hardware.
Check Windows Event Viewer for Claude-related errors:
Open Event Viewer → Windows Logs → Application and look for errors around the time of the timeout. This can reveal if Windows is blocking VM processes.
Verify network access:
If you're on a corporate network or VPN, try disabling them temporarily. The workspace VM connects through our cloud infrastructure and can be blocked by firewalls or proxies.
If the issue continues after these steps, consider reporting it on GitHub Issues with your system specs (ARM64 Snapdragon X, Windows 11), the screenshot you shared, and any Event Viewer logs — this could be ARM64-specific behavior that needs investigation.
Fin AI Agent • AI Agent • 28m
Did that help with the timeout issue?
Fin AI Agent • 28m
We noticed you might have stepped away. We'll close this conversation for now, but if you need additional assistance you can always follow up here.
Fin AI Agent • 18m
No., it is showing again
Sukhen • 8m
New
Since the VM timeout persists on your ARM64 Snapdragon X system after all standard troubleshooting, this appears to be a hardware-specific issue that needs developer attention.
Report this on GitHub:
File a bug at https://github.com/anthropics/claude-code/issues with:
Your screenshot showing the bash timeout errors
System specs: ASUS Vivobook S16, ARM64 Snapdragon X, Windows 11
Steps you've tried (clean reinstall, registry clean, environment variables)
Any Windows Event Viewer logs from around the timeout
The 60-second VM connection timeout is hardcoded in the MCP SDK, and ARM64 Snapdragon X is newer hardware that may need additional startup time or compatibility fixes.

Steps to Reproduce

Version 1.8089.1 (b98a06)

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

No response

Claude Code Version

Version 1.8089.1 (b98a06)

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Windows Terminal

Additional Information

No response