[BUG] Claude Desktop (Windows, MSIX) fails to launch after spontaneous close — "reinstall required" dialog, Repair does not fix it
What's Wrong?
Claude Desktop closed on its own (no user action — no click, no menu command) while in use. Attempting to relaunch immediately produced a Windows package-activation failure dialog:
"無法開啟此應用程式。Claude 發生問題。請從原始安裝位置重新安裝此應用程式,或聯絡您的系統管理員。" (English equivalent: "This app can't open. Claude has a problem. Reinstall this app from its original installation location, or contact your administrator.")
This is a Windows APPX/MSIX activation-failure dialog, not a normal app-crash/error dialog — no crash was recorded (see Diagnostics below).
Using Settings → Apps → Claude → Repair did not resolve it — identical dialog on the next launch attempt. A full system reboot also did not resolve it. During this outage, sending messages via claude.ai in a browser (an independent session, unrelated to the local desktop app) also produced no response for several minutes, though this may be a coincidental, unrelated issue — no incident was listed on status.claude.com for that window.
Reinstalling using the original, already-downloaded installer (no new download) fixed it — the app opened normally afterward. This points to a corrupted/deregistered MSIX package identity under the user profile rather than corrupted application files, consistent with the pattern described in #53247.
What Should Happen?
The app should either recover on its own after an unexpected close, or the in-app "Repair" option (Settings → Apps → Repair) should be able to fix package-registration issues without requiring a full reinstall.
Error Messages/Logs
Windows dialog (zh-TW):
無法開啟此應用程式。Claude 發生問題。請從原始安裝位置重新安裝此應用程式,或聯絡您的系統管理員。
CoworkVMService log — clean stop/restart, no crash trace:
...Persistent RPC: connection ended: failed to read length: EOF
...Session <path> has no active connections, idle timeout starts
...Service stop requested
...Shutdown signal received...
...Server stopped
...Service stopped.
[~21 min gap]
...Starting CoworkVMService as Windows service
...Claude VM Service starting...
...Waiting for configuration from app via 'configure' method...
...Service ready. Listening on \\.\pipe\cowork-vm-service
Warning: failed to configure recovery actions (a crashed service will stay down until reboot): open service: Access is denied.
[this warning is present on every service start, unrelated to this incident]
Windows Application/System event logs for the incident window: no APPCRASH (Event ID 1000) and no AppHang (Event ID 1002) entries for Claude.exe.
Most recent WER AppHang archive for Claude predates the incident by about a day and does not correspond to this event.
Steps to Reproduce
Root cause/trigger unknown — could not be reproduced on demand. Steps as observed:
- Use Claude Desktop normally.
- App closes on its own with no user action.
- Attempt to relaunch → "reinstall required" activation-failure dialog appears immediately.
- Settings → Apps → Claude → Repair → does not fix it; same dialog on next launch.
- Reboot the machine → issue persists.
- Reinstall using the same, already-downloaded installer (no new download) → app launches normally again.
Claude Model
Not applicable / Not sure — this is a Desktop app launch failure, not a model-behavior issue.
Is this a regression?
I don't know
Last Working Version
Not applicable — app had been working normally in the same session shortly before the spontaneous close.
Claude Code Version
Desktop app: 1.34493.1 (Electron 42.9.2)
Bundled Claude Code CLI: 2.1.227
Platform
Anthropic API
Operating System
Windows
Terminal/Shell
Not applicable — this is a Desktop app GUI issue, not a terminal/CLI session.
Additional Information
- Related prior reports on this repo describing the same failure class: #46390 (near-identical symptom, closed as duplicate), #53247 (proposes the orphaned Silo/Job Object root-cause theory this report's evidence is consistent with), #48437, #59471, #49917 (MSIX registration failures with various HRESULTs).
- A screenshot of the exact activation-failure dialog is available and can be attached on request.
- Diagnostics (event logs, WER archive listing, service log) were gathered from a still-functioning Claude Code CLI session on the same machine, after the app was already reinstalled and working again — so they reflect the state shortly after recovery, not a live capture during the outage itself.
- [x] I have searched existing issues and this hasn't been reported yet (closest matches are noted above as related, not exact duplicates)
- [x] This is a single bug report
- [x] I am using the latest version of Claude Code
This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗