[BUG] Cowork workspace fails to start on Linux — Electron int3 trap crash after VM bundle download completes
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?
I know you don't support Linux (shame, Anthropic, shame...), but I'll file this anyway.
The Cowork workspace consistently fails to start with "Request timed out: isGuestConnected". The VM bundle downloads successfully every time, but the VM never actually boots. The Electron process crashes internally before spawning any KVM/QEMU process.
What Should Happen?
Workspace starts and connects successfully after VM bundle downloads.
Actual behaviour:
[14:38:07] rootfs.vhdx.zst checksum validated — saved to cache
[14:38:09] vmlinuz.zst checksum validated — saved to cache
[14:38:14] initrd.zst checksum validated — saved to cache
Log then goes completely silent — no VM:steps, no start_vm, no wait_for_guest entries. No KVM or QEMU process ever spawns.
dmesg shows the Electron process hitting an int3 breakpoint trap at the exact time the VM boot should occur:
Error Messages/Logs
traps: electron[415028] trap int3 ip:5598b952a8ab sp:7fff418b4590 error:0 in electron[72ad8ab,5598b490d000+9705000]
Steps to Reproduce
- Launch Claude Desktop on Linux
- Navigate to Cowork / "Let's knock something off your list"
- App shows "Starting Claude's workspace..." indefinitely
- After ~60 seconds: "Failed to start Claude's workspace — Request timed out: isGuestConnected"
Claude Model
None
Is this a regression?
Yes, this worked in a previous version
Last Working Version
Prior to 1.1.7714...sorry, I didn't capture it.
Claude Code Version
2.1.80 (Claude Code)
Platform
Anthropic API
Operating System
Other Linux
Terminal/Shell
Other
Additional Information
System checks (all passing)
/dev/kvm permissions: crw-rw-rw- (world accessible)
KVM modules loaded: kvm_amd + kvm confirmed via lsmod
virtiofsd installed: yes (Arch extra/virtiofsd 1.13.3-1)
Note: Arch installs virtiofsd to /usr/lib/virtiofsd not /usr/bin/ — symlink created to /usr/bin/virtiofsd
Additional Context
Previous version 1.1.7464 had a different failure: VM timed out on wait_for_guest with virtiofsd socket not ready errors, plus environments-2026-03-01 beta header API rejections blocking Cowork entirely
Version 1.1.7714 fixed the beta header issue but introduced the int3 crash
The new version also changed the VM bundle directory structure from flat claudevm.bundle/ to claudevm.bundle/wvm-<random>/ — unclear if the boot code was updated to match
This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗