Thank you!! Yes - that's the goal of this project. We're trying to stop agents from taking unsafe actions. You can either block actions or tag them for human approval