[BUG] Cowork: Failed to create bridge sockets after 5 attempts on every task
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 fails on every new task — including trivial ones like "what is 2+2" — with the error:
Error: Failed to create bridge sockets after 5 attempts
Claude Code process exited with code 1
The crash happens immediately after process spawn, with a duration of ~900–1300ms. It affects both manually started tasks and scheduled tasks (my weekly scheduled task has been failing in the background as well).
Related to #23746 and #19503, which describe the same error.
Log excerpt from ~/Library/Logs/Claude/cowork_vm_node.log:
2026-04-16 14:23:05 [info] [Spawn:create] id=d26112e8-d6de-403c-99f7-1ae26f2b133c name=tender-kind-hypatia cmd=/usr/local/bin/claude args=--output-format stream-json --verbose --input-format stream-json ...
2026-04-16 14:23:05 [info] [Spawn:vm] id=d26112e8-d6de-403c-99f7-1ae26f2b133c OAuth token approved with MITM proxy
2026-04-16 14:23:05 [info] [Process:d26112e8-d6de-403c-99f7-1ae26f2b133c] Spawn confirmed, flushing 2 buffered stdin chunks
2026-04-16 14:23:05 [info] [Spawn:vm] id=d26112e8-d6de-403c-99f7-1ae26f2b133c Spawn succeeded in 174ms
2026-04-16 14:23:06 [warn] [vm-stderr d26112e8] Error: Failed to create bridge sockets after 5 attempts
2026-04-16 14:23:06 [info] [Process:d26112e8-d6de-403c-99f7-1ae26f2b133c] Exited, code=1, signal=null, duration=1268ms, oom=false
VM startup itself succeeds — network is CONNECTED, API reachability is REACHABLE, SDK installs fine. The failure is specifically in the bridge socket creation step inside the VM, right after Claude Code spawns.
One earlier scheduled run also failed with a different but likely related error:
API Error: 587 {"error":"coworkd_upstream_failed","category":"dns_failure","detail":"dial tcp: lookup api.anthropic.com on 172.16.10.1:53: no such host"}
Not sure if that's the same root cause or a separate VM networking issue.
What Should Happen?
Cowork tasks should execute successfully. A new task with a trivial prompt like "what is 2+2" should return a response, not crash with "Claude Code process exited with code 1".
Error Messages/Logs
Error surfaced in UI:
Claude Code process exited with code 1
Underlying error from ~/Library/Logs/Claude/cowork_vm_node.log:
2026-04-16 14:23:05 [info] [Spawn:create] id=d26112e8-d6de-403c-99f7-1ae26f2b133c name=tender-kind-hypatia cmd=/usr/local/bin/claude args=--output-format stream-json --verbose --input-format stream-json --max-thinking-tokens 31999 --effort medium --model claude-opus-4-7 ...
2026-04-16 14:23:05 [info] [Spawn:vm] id=d26112e8-d6de-403c-99f7-1ae26f2b133c OAuth token approved with MITM proxy
2026-04-16 14:23:05 [info] [Process:d26112e8-d6de-403c-99f7-1ae26f2b133c] Spawn confirmed, flushing 2 buffered stdin chunks
2026-04-16 14:23:05 [info] [Spawn:vm] id=d26112e8-d6de-403c-99f7-1ae26f2b133c Spawn succeeded in 174ms
2026-04-16 14:23:06 [warn] [vm-stderr d26112e8] Error: Failed to create bridge sockets after 5 attempts
2026-04-16 14:23:06 [info] [Process:d26112e8-d6de-403c-99f7-1ae26f2b133c] Exited, code=1, signal=null, duration=1268ms, oom=false
2026-04-16 14:23:06 [info] [Process:d26112e8-d6de-403c-99f7-1ae26f2b133c] Cleaned up, remaining active=0
VM startup log (completes successfully before the bridge socket failure):
2026-04-16 14:23:02 [info] [VM] Network status: CONNECTED
2026-04-16 14:23:04 [info] [VM] API reachability: REACHABLE
2026-04-16 14:23:04 [info] [VM:start] Startup complete, total time: 9171ms
Steps to Reproduce
- Open Claude Desktop
- Switch to Cowork
- Start a new task with any prompt (e.g. "what is 2+2")
- Task immediately fails with "Something went wrong — Claude Code process exited with code 1"
- Same error on every retry, every new task, regardless of prompt content
Claude Model
Not sure / Multiple models
Is this a regression?
Yes, this worked in a previous version
Last Working Version
_No response_
Claude Code Version
Claude 1.3036.0 (cf521c) 2026-04-16T16:17:33.000Z
Platform
Anthropic API
Operating System
macOS
Terminal/Shell
Terminal.app (macOS)
Additional Information
What I tried:
- Reinstalled Claude Desktop entirely
- Deleted the Cowork VM bundle (rm -rf ~/Library/Application Support/Claude/vm_bundles) and let it re-download/rebuild
- Verified TMPDIR is set and writable
- Installed socat on host via Homebrew (brew install socat)
- Symlinked /opt/homebrew/bin/socat → /usr/local/bin/socat
- Fully quit and relaunched Claude between each attempt
- Switched from Legacy Model to Opus 4.7
None of these resolved the bridge socket creation failure. The error appears to originate inside the VM sandbox, not on the host.
This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗