[BUG] Cowork Windows — remote-bash reports "no Plan9 drive shares mounted" after Sept 2026 Windows update; host registers 7 shares successfully

Status Open
Maintainer reply None cached
Activity 4 comments · opened Sep 9, 2026

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?

What's Wrong?

Cowork's shell to the local machine fails with:

no Plan9 drive shares mounted

File read/write through the mounted folder still works. Only script
execution on the host is blocked — the session falls back to the cloud
sandbox. Onset was immediately after a Windows update on 2026-09-08.

The host service registers all shares without error, but remote-bash
receives none.

Environment

  • Claude Desktop 1.49585.0 (MSIX, Claude_pzs8sxrjxfjjc)
  • Windows 11, 32 GB RAM
  • Recent updates: KB5124008 (9/9), KB5124007 (9/8), KB5126052 (9/8)
  • Working folder: C:\Users\cronm\Desktop\Vault\Music (local NTFS, not

OneDrive, no spaces)

Evidence

Host side succeeds — C:\ProgramData\Claude\Logs\cowork-service.log:

[VM] User token set for Plan9 identity
[VM] Plan9 share added: c -> C:\ (port=9902, readOnly=false)
[VM] Plan9 share added: d -> D:\ (port=9903, readOnly=false)
[VM] Plan9 share added: e -> E:\ (port=9904, readOnly=false)
[VM] Plan9 share added: g -> G:\ (port=9906, readOnly=false)
[VM] Plan9 share added: h -> H:\ (port=9907, readOnly=false)
[VM] Plan9 share added: i -> I:\ (port=9908, readOnly=false)
[VM] Plan9 share added: j -> J:\ (port=9909, readOnly=false)
[VM] Sent plan9Shares notification (7 shares)

No Failed to add Plan9 share and no 0x80070005 anywhere in the log —
this is not the access-denied failure in #43896 or #44486.

App side, %LOCALAPPDATA%\Claude\logs\main.log:

[remote-bash] user=rcw-... mounts=Music:rw cmdLen=60

Reproduction matrix

Two external USB drives (J: WD Elements, G: Seagate Backup Plus) are
mapped as shares along with the internal drives. Both conditions must
hold for the shell to work:

| Windows update | J and G attached | Result |
| --- | --- | --- |
| Installed | Attached | Fails |
| Installed | Unplugged | Fails |
| Uninstalled | Attached | Fails |
| Uninstalled | Unplugged | Works |

Uninstalling the update and reinstalling it toggles the failure
reproducibly.

Tried, no effect

  • Reboot (multiple)
  • Restart-Service CoworkVMService
  • Full uninstall/reinstall of Claude Desktop
  • Fresh Cowork sessions
  • icacls grant to NT VIRTUAL MACHINE\Virtual Machines (access denied;

and logs show no permission failure to fix)

Expected

Shares registered successfully by the service should be visible to
remote-bash. If a drive cannot be shared, that drive should be skipped
with a warning rather than the whole mount set being unavailable.

Open question

Three updates installed across 9/8–9/9; the rollback removed the most
recent. Which specific KB triggers this is not established.

Related issues to link in the body: #44486, #43896, #28890.

What Should Happen?

Shares registered successfully by the service should be visible to
remote-bash. If a drive cannot be shared, that drive should be skipped
with a warning rather than the whole mount set being unavailable.

Error Messages/Logs

no Plan9 drive shares mounted

[VM] User token set for Plan9 identity
[VM] Plan9 share added: c -> C:\ (port=9902, readOnly=false)
[VM] Plan9 share added: d -> D:\ (port=9903, readOnly=false)
[VM] Plan9 share added: e -> E:\ (port=9904, readOnly=false)
[VM] Plan9 share added: g -> G:\ (port=9906, readOnly=false)
[VM] Plan9 share added: h -> H:\ (port=9907, readOnly=false)
[VM] Plan9 share added: i -> I:\ (port=9908, readOnly=false)
[VM] Plan9 share added: j -> J:\ (port=9909, readOnly=false)
[VM] Sent plan9Shares notification (7 shares)

[remote-bash] user=rcw-... mounts=Music:rw cmdLen=60

no Failed to add Plan9 share and no 0x80070005 appear anywhere in cowork-service.log.

Steps to Reproduce

  1. Windows 11 machine with Claude Desktop 1.49585.0 (MSIX) and two

external USB drives attached (J: WD Elements, G: Seagate Backup Plus),
alongside internal drives C, D, E, H, I.

  1. Install the Sept 2026 Windows updates (KB5124008, KB5124007, KB5126052).
  2. Open Cowork and add a local folder — C:\Users\cronm\Desktop\Vault\Music.
  3. Ask Cowork to run any script on the host, e.g. a Python file in the

added folder.

  1. Observe: file read/write works, script execution fails with

"no Plan9 drive shares mounted".

  1. Check C:\ProgramData\Claude\Logs\cowork-service.log — all 7 shares

registered successfully, no errors.

  1. Uninstall the Windows update AND detach both USB drives, then reboot.

Scripts now run on the host.

  1. Restore either condition (reinstall the update, or reattach the

drives) and the failure returns.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

1.49585.0.0

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

PowerShell

Additional Information

_No response_

View original on GitHub ↗

3 Comments

alignedactionconsult-svg · 3 days ago

Independent repro, and I can narrow the trigger to KB5124008 specifically.

Environment

  • Claude Desktop 1.49585.0 (41ad1d) — same build as OP — MSIX (Claude_pzs8sxrjxfjjc)
  • Windows 11 Pro 25H2, build 26200 (→ 26200.9445 after KB5124008)
  • No external USB drives attached

Symptom
Shell worked on 9/8. From the first session on 9/9, every bash call failed — file tools kept working. Error, verbatim:

bash failed on resume, create, and re-resume.
resume: RPC error -1: failed to mount /mnt/.virtiofs-root/shared/c/Users/<user>/AppData/Local/Packages/Claude_pzs8sxrjxfjjc/LocalCache/Roaming/Claude/local-agent-mode-sessions/<...>/outputs as outputs: source path <...> is under Plan9 share "c" which is not mounted
create: RPC error -1: ensure user: user <session-name> already exists unexpectedly: uid=1007 gid=1007

Slightly different wording from the OP ("no Plan9 drive shares mounted") but same mechanism: the VM never sees the c share. Note the second error — the sandbox's per-session user survived three Windows restarts, so resume and create were both blocked until the share came back.

Update timeline (Get-HotFix)

  • KB5124007 (SSU) — installed 9/8
  • KB5124008 (Sept CU, 26200.9445) — installed 9/9
  • KB5126052 (.NET Framework) — installed 9/9

Isolation
Uninstalled only KB5124008 and restarted. Shell mounted immediately in a new session, and an already-stuck existing session recovered on its next bash call without restarting the app. KB5124007 and KB5126052 remain installed (confirmed via Get-HotFix). So KB5124008 is the trigger, and external drives aren't a required co-factor.

Tried before the rollback, no effect

  • Fully quitting/restarting the app; 3× Windows restart (Restart, not Shut down)
  • Fresh Cowork sessions (new session name, same mount failure)
  • CoworkVMService, vmcompute, hvhost all Running; VirtualMachinePlatform Enabled
  • Administrator Protection confirmed off (TypeOfAdminApprovalMode = 1)

Updates are paused for now. Happy to provide cowork-service.log if useful.

grodinsm · 3 days ago

confirming same issue here, same updates.

What I tried: in Powershell (Admin) I ran 'wsl -- update' and it restored the ability to start the sandbox, but it still couldn't load the virtofs mounts.

techfury90 · 3 days ago

Confirming here too- we had to back out KB5124008 to fix a coworker's scheduled data conversion task.

Showing cached comments. Read the full discussion on GitHub ↗