See here:
https://github.com/CesiumGS/cesium-native/blob/v0.56.0/extern/vcpkg/ports/ktx/portfile.cmake#L3
This isn't catastrophic, but it's poor practice (on my part). I did it because I was actively pushing changes into the upstream project at the time, and so I expected the need for the fork to only be temporary. For various reasons, that has taken longer than expected, and may take awhile longer still. So we may want to move to a CesiumGS fork, at least.
For reference, here's the PR that I opened upstream:
KhronosGroup/KTX-Software#1075