Tensorflow Agents with actual environment instead of python class environment

Question

As per Wikipedia on Reinforcement Learning

Two elements make reinforcement learning powerful: the use of samples to optimize performance and the use of funct