Adding an unused tensor to the model created with the Keras functional API (TensorFlow 2.0)

I have some variables I would like to save in a Keras model so that when the model is saved to disk and later loaded, the variables are accessible from the model. However, the model itself doesn’t use these variables: they are intended as hints to the user about what the model does and how to use it.

With the Keras functional API, I don’t see an obvious way to do this. I could add a layer whose only function is to store these variables, but I think I would still need to connect it to the rest of the graph somehow so that it becomes part of the model. Is there a simple way to do this that I am missing, or do I need to use a different API or write these variables to a separate file?