Confidence measures and feature importance for RF

Hello all,

For Random Forests using XGBoost:
Anyone can point me to show to get each individual tree? preferably as a function that I can pass a vector of features to?

I am working on a measure of confidence for the RF and I would like to see what each tree has to say.

Also I would like to ask how to compute the Gini importance (or purity)? I am aware there are others but I am interested in this one in particular. Seems to me a bit easier to explain and understand than SHAP for example.

I am using the standalone XGBoost python API but I would appreciate any comment for any language or even the theory behind.

I would like to be able to see this step by step by myself and not with a premade method. I also don’t need plotting , I just want to see the features with a value associated to it showing a relative importance score.

If anyone knows or can tell me about how confident a forest is on a classification I would appreciate it