How are you learning agent pen testing?

Exactly the title. The traditional app sec pen testing and pen testing an AI agent are different things. I know the underlying vulnerability is still same but the way you attack and get it exposed are different. Example: Social Engineering. You need to be good at that to be able to test properly. I am just curious, how teams are up skilling? Any tools you are using that assist you in testing or something else? submitted by <a