Power Platform Community Forum Thread Details

I'm automating a Citrix-hosted application using Power Automate Desktop (PAD) across two virtual machines and running into selector instability.
My Setup:
VM A: Where I design and orchestrate desktop flows
VM B: Runs the Citrix application (hosted app I want to automate)
Power Automate Agent for virtual machines is installed and connected on VM B
Since Citrix doesn't expose UI elements directly, I'm using image-based and keyboard automation in the PAD flow triggered on VM B
Issue:
Every time I run the flow, I get:
"Couldn't test the selector, the parent screen was not found."
This happens even when interacting with File Explorer, Downloads window, or standard Windows dialogs
It only works if I re-capture selectors every time, which isn’t reliable
Also facing issues launching browsers in VM B for any post-processing
Selectors break even with minor changes in screen resolution or window state
What I’ve Tried:
Set fixed resolution and scaling in VM B
Added delays, image-based clicks, and keyboard shortcuts
PAD agent is installed correctly and machine is connected
Looking For:
A stable method to automate Citrix applications using PAD and PAD Agent
Best practices for image/keyboard automation within Citrix windows
Ways to avoid constant re-capturing of selectors or prevent breakage
Alternatives or enhancements that could improve reliability of this setup

Categories:

Power Automate Desktop

@VM-23070629-0

You're facing a common challenge when automating Citrix-hosted applications using Power Automate Desktop (PAD) — selector instability and resolution sensitivity. Here's a comprehensive guide to help stabilize your automation and improve reliability:

✅ Understanding the Challenge

Citrix applications render UI elements as bitmaps, not native Windows controls. This means:

Selectors often fail because PAD can't "see" the UI elements.
Image-based automation becomes the fallback, but it's sensitive to resolution, scaling, and window state.

🛠️ Best Practices for Stable Citrix Automation

1. Use Image and OCR Automation Strategically

Use high-quality, consistent screenshots for image matching.
Prefer OCR-based text recognition for dynamic elements (e.g., buttons, labels).
Use relative coordinates when clicking near recognized text.

2. Standardize VM B Environment

Set fixed resolution (e.g., 1920x1080) and 100% scaling.
Disable auto-scaling and DPI awareness in Citrix settings.
Ensure the Citrix app always opens in the same position and size.

3. Avoid Selectors for Citrix UI

Use selectors only for native Windows dialogs (e.g., File Explorer).
For Citrix windows, rely on image-based actions and keyboard shortcuts.

4. Use Wait and Retry Logic

Add “Wait for image/text” actions before clicking.
Use “Retry” loops to handle slow Citrix rendering.

5. Launch Browsers via Command Line

If launching browsers fails on VM B:

Start-Process "C:\Program Files\Google\Chrome\Application\chrome.exe" "https://your-url.com"

Use PAD’s “Run application” action with full path and arguments.

🔄 Alternative Approaches

✅ Use Citrix Virtual Channel SDK

If you have control over the Citrix environment, consider using the Citrix Virtual Channel SDK to expose more automation-friendly interfaces.

✅ Hybrid Automation

Split the flow:

Use PAD on VM A for orchestration.
Use PowerShell or remote scripts on VM B for launching apps or post-processing.

🧪 Debugging Tips

Use PAD’s “Take Screenshot” action to verify what VM B sees.
Log all actions and failures to a file for traceability.
Test flows with minimal UI changes to isolate instability causes.

🏷️ Tag me if you have any further questions or if the issue persists.

✅ Click "Accept as Solution" if my post helped resolve your issue—it helps others facing similar problems.

❤️ Give it a Like if you found the approach useful in any way.

Community site session details