[BUG] 1.3036.0 (cf521c): All Cowork sessions crash with "exited with code 1" on macOS - Desktop-to-SDK spawn failure, SDK 2.1.111
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?
After updating to Claude Desktop 1.3036.0 (cf521c) on 2026-04-16, every Cowork session crashes with "Claude Code process exited with code 1" approximately 3 seconds after starting. The crash occurs after the VM instance is assigned and the session transitions to running, but before any first response. The Claude Code process dies silently with no stderr output, no CSP violations, and no VM boot errors visible in main.log.
Timing correlates with the Opus 4.7 release earlier today, though the crash reproduces across all models (Opus 4.7, Opus 4.6, Sonnet 4.6), so it's not model-specific.
What Should Happen?
Cowork sessions should start and respond to messages.
Error Messages/Logs
From ~/Library/Logs/Claude/main.log:
2026-04-16 14:45:10 [info] [LocalAgentModeSessionManager] SDK version 2.1.111 will be used
2026-04-16 14:45:10 [info] Using Claude VM spawn function for session
2026-04-16 14:45:10 [info] [Lifecycle] Session local_38cf1389...: initializing → running
2026-04-16 14:45:11 [error] Session local_38cf1389... query error: Claude Code process exited with code 1
2026-04-16 14:45:11 [info] [Lifecycle] Session local_38cf1389...: running → idle
2026-04-16 14:45:11 [info] [CycleHealth] Unhealthy cycle: {
session_id: 'local_38cf1389-ef67-4b00-be9e-b9ad6c431520',
vm_instance_id: '699449a1-66d4-4725-b31a-6256cf3df68c',
model: 'claude-opus-4-7[1m]',
cycle_health: 'unhealthy',
had_first_response: false,
seconds_to_outcome: 3,
is_resume: false,
unhealthy_reason: 'system_error',
error_message: 'Claude Code process exited with code 1',
error_category: 'process_crashed'
}Steps to Reproduce
Update to Claude Desktop 1.3036.0 (cf521c) on macOS
Open Cowork
Start any new session (or resume an existing one) with any model
Observe crash within ~3 seconds
Claude Model
Opus
Is this a regression?
Yes, this worked in a previous version
Last Working Version
_No response_
Claude Code Version
2.1.86 (Claude Code)
Platform
Anthropic API
Operating System
macOS
Terminal/Shell
Terminal.app (macOS)
Additional Information
Troubleshooting already attempted (no effect):
Full quit and relaunch
Sign out and sign back in
Help -> Troubleshooting -> Clear Cache and Restart
Reproduces across Opus 4.7, Opus 4.6, and Sonnet 4.6
Reproduces on both new and resumed sessions
21 Comments
Duplicate of this: https://github.com/anthropics/claude-code/issues/49358
Having same issue:
Claude Code version: 2.1.31
macOS version: 26.3.1
The exact error: “Claude Code process exited with code 1”
Confirming I'm hitting the same bug, also today (2026-04-16), also on macOS, also immediately after the Claude Desktop update.
Same fingerprint in my logs:
SDK version 2.1.111
Model: claude-opus-4-7
"Using Claude VM spawn function for session"
Lifecycle: initializing → running → Claude Code process exited with code 1
had_first_response: false, seconds_to_outcome: 2, unhealthy_reason: system_error, error_category: process_crashed
Relevant log excerpt:
2026-04-16 11:48:55 [info] [LocalAgentModeSessionManager] SDK version 2.1.111 will be used
2026-04-16 11:48:55 [info] Using Claude VM spawn function for session
2026-04-16 11:48:55 [info] [Lifecycle] Session local_10373bc7...: initializing → running
2026-04-16 11:48:56 [error] Session local_10373bc7... query error: Claude Code process exited with code 1
2026-04-16 11:48:56 [info] [CycleHealth] Unhealthy cycle: {
model: 'claude-opus-4-7',
cycle_health: 'unhealthy',
had_first_response: false,
seconds_to_outcome: 2,
unhealthy_reason: 'system_error',
error_message: 'Claude Code process exited with code 1',
error_category: 'process_crashed'
}
Also tried: full quit + relaunch, signing out and back in, checking for pending terms on claude.ai (nothing to accept) — no effect. Reproduces on every session immediately after the update, so the timing and SDK 2.1.111 correlation you called out looks right.
Confirming I am having the same issue after relaunching Claude with the newest version
Claude desktop version: 1.3036.0 (cf521c)
macOS version: Tahoe 26.3.1 (25D2128)
Exact error: “Claude Code process exited with code 1”
same issue after relaunching with newest version
Product: Claude Code Desktop (Cowork tab)
OS: macOS (Apple Silicon M1)
Desktop app version: Version 1.3036.0 (cf521c)
Exact error: “Claude Code process exited with code 1”
What changed: it worked before the latest update, and only Cowork is affected (Chat and Code still work)
Steps to reproduce: open Desktop app → go to Cowork tab → submit a prompt → see the error
Extra context: /doctor output shows native install v2.1.111 and no duplicate installs
same issue here after relaunching with newest version
Product: Claude Code Desktop
OS: Tahoe 26.3
Exact error: “Claude Code process exited with code 1”
Same issue here after the recent Cowork update. Session ID: b70779cf-f783-4675-b265-52f11b54b4af — Haiku subagent spawned internally by Cowork crashes on tool_reference blocks (Windows 11, cc_version 2.1.111.c9e).
Same issue here after relaunching with the newest version
Product: Claude Code Desktop
OS: Tahoe 26.5
Exact error: “Claude Code process exited with code 1”
Also having this issue,
App version: 1.3036.0
macOS version: Tahoe 26.2
Error: Failed to create bridge sockets after 5 attempts from ~/Library/Logs/Claude/cowork_vm_node.log
<img width="379" height="178" alt="Image" src="https://github.com/user-attachments/assets/e6cfed66-4d18-4b7d-88d2-23f83e05b6a5" />
same issue here after relaunching with the newest version
<img width="664" height="157" alt="Image" src="https://github.com/user-attachments/assets/3d4d2c2b-845e-47bf-b619-7521e7da3853" />
this seems to be an issue when trying to switch to opus 4.7 from within existing projects. using Opus 4.6 seems to work.
Same issue. I just downloaded Claude for the first time and using Cowork for the first time and get this message:
<img width="479" height="274" alt="Image" src="https://github.com/user-attachments/assets/e752d353-1234-4635-b224-15d9078300ad" />
Claude Code Desktop
Windows 11 Pro Version 25H2
OS build 26200.7623
Experience Windows Feature Experience Pack 1000.26100.275.0
It happens using Opus 4.7 and 4.6
This isn't the case for me. Both 4.6 and 4.7 do not work and receive the same error message. Even if there is no exisiting project selected
Me too!
Same here
Claude Desktop: 1.3036.0 (cf521c), built 2026-04-16T16:17:33.000Z
Claude Code binary: 2.1.111
macOS (Apple Silicon)
Same crash signature — binary passes direct invocation (--version and -p "hello" both succeed), OAuth token valid, crash occurs in Using Claude VM spawn function for session path at ~3s with had_first_response: false.
Ruled out: clearing claude-code-vm state, clearing caches, vm_bundles removal.
Log excerpt:
[info] [LocalAgentModeSessionManager] SDK version 2.1.111 will be used
[info] Using Claude VM spawn function for session
[error] Session local_... query error: Claude Code process exited with code 1
[info] [CycleHealth] Unhealthy cycle: { seconds_to_outcome: 3, had_first_response: false, error_category: 'process_crashed' }
CLI works fine as a workaround.
ongoing status of glitch from Anthropic: https://status.claude.com/#
I new i shouldnt have clicked update :/
Also Haiku doesn't work. Luckily Claude Code does work, just Cowork seems to be affected
v1.3109.0 release worked successfully for me
This is a duplicate of #49358, which was fixed as of version 1.3109.0.
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.