[feature]: Ability to open multiple windows

[philRVT]

What would you like

I would like the ability to open multiple Claude Desktop windows. Not split view; full Claude Desktop windows. Supposedly this can be done in the native Windows and Mac Claude Desktop applications.

Use case

Claude Desktop Debian allows me to use the "Open In > Split View" for two Claude Code forks in my Ubuntu 24 / Gnome 46 desktop.

The problem is the notifications: if one of my Claude Code split views sends a notification -- and I click on it -- then the Claude Code split view disappears and all I see is the forked instance that sent the notification.

...which means that if I had partially-typed a prompt out in the split view that went poof -- and never sent that prompt -- what I had typed goes poof as well.

Thank you so much for a great application to let me use Claude Desktop in my Ubuntu environment!

Existing workarounds

None that I can think of -- other than to go back to terminal mode (please no...the prompt editor is really bad in comparison)

[github-actions]

Automated draft — AI analysis, not maintainer judgment. This bot looked at the issue but couldn't reach a confident read. Routing to a human for review.

Reason: ambiguous bug/enhancement classification

https://github.com/aaddrick/claude-desktop-debian/actions/runs/26310851712 has the raw classification artifact if helpful for context.

(This bot processes issue text via Anthropic's API. See README §Privacy for what that means.)

[aaddrick]

Hey! I get why this would be useful. Splitting conversations across workspaces or monitors is a real workflow, and having a notification yank you out of context is frustrating.

That said, the single-instance lock is enforced upstream by Anthropic on all platforms. Windows and macOS have the same limitation. This project repackages Claude Desktop for Linux, but we don't add features that don't exist upstream. Our scope is Linux parity: if something works on macOS/Windows but breaks on Linux, that's our lane. Multi-window support would require fundamentally modifying the app's architecture, which is well outside what we do here.

If this is something you'd like to see, I'd suggest filing a feature request with Anthropic directly. They control the upstream app and the single-instance decision.

Closing this as out of scope for the project. Thanks for taking the time to open it though.

---

Written by Claude Opus 4.7 via Claude Code