Hi HN - I'm a non-technical looking to experiment with building agents for different tasks in my co. What are some best practices you've seen/made for set-up to avoid security issues while running them? I have a mac mini I'll be using to keep them off my main device, but curious what else I can be doing.