[BUG] Cowork: RPC error 'ensure user already exists unexpectedly' blocks all tasks

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: "Claude Code process exited with code 1" followed by RPC error on all tasks
Started today (2026-04-16) after the Opus 4.6 outage window. Every Cowork task across every project fails immediately with "Claude Code process exited with code 1." When clearing Application Support and relaunching, workspace rebuild fails with: RPC error: ensure user: user amazing-compassionate-thompson already exists unexpectedly: uid=1001 gid=1001
Regular chat works normally. This is Cowork-specific.
Environment: macOS, Claude Desktop 1.3036.0 (cf521c), 2026-04-16T16:17:33.000Z
Tried: full quit/relaunch, logout/login, clearing Caches, moving Application Support aside (restored from backup after rebuild failed with the RPC error above).
Appears related to issues #24190, #24483, #24549, #26129.

What Should Happen?

Cowork should successfully provision the sandbox session and allow tasks to execute normally. Opening any Cowork project and running any task should start without error, whether the session is new or resumed from an existing project.
The RPC "ensure user" step should handle pre-existing session identities gracefully instead of failing with "already exists unexpectedly." Regular chat continues to work, so authentication and account state are fine. Only the Cowork VM provisioning is failing.

Error Messages/Logs

Claude Code process exited with code 1
You can restart the conversation from an earlier message.

RPC error: ensure user: user amazing-compassionate-thompson already exists unexpectedly: uid=1001 gid=1001

The first error appears on any task attempt in an existing Cowork project. The second error appeared after clearing ~/Library/Application Support/Claude and relaunching, during the workspace rebuild step.

Steps to Reproduce

1. Open Claude Desktop (version 1.3036.0, macOS).
2. Open any Cowork project.
3. Send any prompt or attempt any task.
4. Error appears immediately: "Claude Code process exited with code 1."
5. Reproduces across every Cowork project. Regular chat mode works normally.

Additional reproduction after local cleanup attempt:

1. Fully quit Claude.
2. Move ~/Library/Application Support/Claude aside to force a fresh rebuild.
3. Relaunch Claude and log in.
4. Workspace rebuild begins, then fails with: "RPC error: ensure user: user amazing-compassionate-thompson already exists unexpectedly: uid=1001 gid=1001"

Not reproduced: the issue is consistent; no intermittent successes.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

the version before Claude 1.3036.0 (cf521c) 2026-04-16T16:17:33.000Z

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

Issue began today (2026-04-16), the same day as the Opus 4.6 outage that resolved at 00:26 PT / 7:26 UTC per status.claude.com. Claude Desktop auto-updated to 1.3036.0 (cf521c, build timestamp 2026-04-16T16:17:33.000Z) at some point during or after this window. Unclear whether the bug is introduced by the 1.3036.0 build itself or is residual state from the outage.
Error pattern ("ensure user" RPC failures during Cowork session provisioning) matches multiple prior reports: #24190, #24483, #24549, #26129. This instance differs in the specific failure mode: "already exists unexpectedly" (collision) rather than "should already exist but does not" (missing). Both appear to be failure modes of the same provisioning path.
Regular chat works normally throughout. Account, auth, and API connectivity are not implicated.