eve looks like a very promising agent framework.

Built-in： - Durable execution - Sandboxed compute - Human-in-the-loop approvals - Subagents - Evals - and more

I like the emphasis on evals right away.

Should I do a tutorial on it？