The only reference we have for TreeShap in H2O (GBM, XGBoost) is in a blog post: https://www.h2o.ai/blog/h2o-release-3-26-yau/. It would be great to have an official reference for this feature.
Let's add it somewhere to the User Guide. We have a demo here we can use for a code example: https://github.com/h2oai/h2o-3/blob/master/h2o-py/demos/shap_values_drf.ipynb
Another code example here: https://sefiks.com/2019/10/10/interpretable-machine-learning-with-h2o-and-shap/