Hi Leon Xu,
I can understand why you're investigating this as a possible Shell issue. Based on your description, Windows Peek is consistently identifying the correct Electron window, which makes it reasonable to suspect that the failure is happening at the activation or foreground-transition stage rather than during thumbnail resolution itself. However, there's a few things I need to clarify upfront here so you can have correct expectations going forwards and find the answers you need:
- Microsoft Q&A is just a community support forum. The contributors here, including moderators and external staff, are fellow users, who share suggestions for technical issues based on official documents and their experience.
- Your issue gets into internal Explorer/taskbar implementation details that are not publicly documented. Microsoft does not generally publish the exact activation algorithm used by the Windows 11 taskbar, nor the internal ETW event mappings for every foreground transition path.
- Because of that, while posting here may help attract others who observed similar behavior or have the same queries, it is unlikely that the thread itself will be reviewed by the responsible engineering team. Realistically, I don't think anyone here can confirm your testing, answer your questions, or escalate these for further investigation.
- Given those limitations, consider reporting this through Microsoft’s official channels:
- Press Win + F to open Feedback Hub > submit under relevant category.
- I recommend including the same information you provided above.
- Also attach any logs, screenshots, videos on this (if possible).
- Give a TL;DR on the issue for better visibility!
You can also open a support case with Microsoft Support, but this is generally not within their scope. For internal diagnostics, an escalation through a developer support channel would likely be required. I do have some thoughts, but note that this is just my assessment from the information provided, not to be confused with confirmation:
- The shared AppUserModelID design itself appears valid and supported.
- The issue does not currently look like proof of an AppUserModelID limitation.
- Since it reproduces on only one machine, I would lean toward either:
- A machine-specific Shell / focus issue or a third-party focus interceptor.
- An application-level focus race occurring after Explorer activates the selected window.
Overall, I would recommend against assigning separate AppUserModelIDs solely as a workaround unless there's a confirmation that cross-process grouping is the trigger. But that's unfortunately not something you likely to get from a public community forum like Microsoft Q&A. I hope you can get better answers from professionals on this.
If you think the answer is helpful to you, please mark this as accepted. If you have extra questions about this answer, feel free to click "Add Comment" and ask!
Note: Follow the steps in our documentation if you want to enable and receive the related email notifications for this thread.