This is a suite of reinforcement learning environments illustrating various safety properties of intelligent agents. - View it on GitHub