The llama-server local-model engine ships in Windows builds only for now.
This placeholder keeps the resource path present on other platforms.
