3 Answers, 1 is accepted
Hello, Neil,
Starting with the 2025 Q1 release, Telerik UI for WPF requires activation through a license key. The new licensing model applies to all Telerik products, not just Telerik UI for Wpf. To activate a NuGet-based project successfully, you must do two things: install the Telerik.Licensing package and make the license key available to the account or environment that builds the application. If either part is missing, the build can log errors and warnings and the application can show watermarks or banners at run time - this is expected.
More information about how to activate Telerik UI for WPF license key is available here: WPF Activating Your License Installing License Key - Telerik UI for WPF
If you have already activate your license key, but still facing some licensing related messages, it would be great if you can share more details about these messages or provide the build output. This information would be useful for us to make further investigation.
Regards,
Nadya | Tech Support Engineer
Progress Telerik
Unlock smarter data exploration in RadGridView with AI-powered features. Enable natural language queries and help users uncover insights faster.
The boilerplate answer does not help with the scenario I specifically mentioned in my original post - agentic AI coding. For others in the same boat the following instructions might help depending on your setup.
## Throwaway Telerik probe apps **A throwaway app that references the Telerik assemblies needs `Telerik.Licensing`, or it will quietly give wrong answers.** Add `<PackageReference Include="Telerik.Licensing" Version="1.9.0" PrivateAssets="All" />` to the probe project. It finds the key already sitting at `%APPDATA%\Telerik\telerik-license.txt`, so nothing needs copying in and no key goes near the project. Without it the controls still construct and behave normally, which is what makes this expensive. A few seconds after the window opens, a "License key missing for Telerik UI for WPF" banner appears over it — and the banner is hit-testable, so it silently swallows clicks aimed at whatever is underneath. A probe driving clicks then reports that nothing happened, which reads exactly like the behaviour under test being absent. The delay is the sting in it: a quick probe finishes before the banner shows and passes, a slower one fails, so the same code gives different answers on different runs. The general lesson is worth more than the specific fix. When a probe reports a negative result, capture a frame of its window before believing it — a screenshot showed the banner immediately, after three runs of theorising about mouse capture, DPI scaling and locked desktops had got nowhere. It is also why a probe should always run a control case it expects to pass: a run where the control fails is a broken harness, not a finding.
That’s a useful detail about the probe apps. The delayed banner and the fact that it can swallow clicks would definitely make agent testing look like the control itself is broken. Capturing a screenshot before drawing conclusions seems like a simple but important step here.

Hi Neil,
Thank you for the provided details. Could you please specify which Telerik product you are using in the app? I am asking as now the forum is opened without setting a specific product. Is the question asked related to a concrete product?