roborev

Thanks! What tools have you been experimenting with? Agreed. That this evolution pushes much of the work into describing desired outcomes and giving sufficient context. To your questions: Emdash helps reduce the setup cost of each environment by allowing you to open an isolated git worktree, copying over env variables and other desired context. And then preserving your conversation per task. That said, you still need to write clear goals and point it in the right direction. I think it's less about team scale and more about individual throughput. My working mode is that I'm actively working on one or two tasks, switching between them as one runs. Then I have a long list of open tasks in the sidebar that are more explorative, quick reviews, issue creation etc. So for me it's not about one-shotting tasks, but instead about navigating between

Adverserial work (be it agent or human). The one difference between "can do" and "should be trusted to do" is the ability to systematically prove that "can do" holds up close to 100% of task instances and under adverserial conditions. Hacking and pentesting are already scaling fully autonomously - and systematically. For now, lower level targets aren't yet attractive as such scale requires sophisticated (state) actors, but that is going to change. So building systems that white-hat prove your code is not only functional but competent are going to be critical not to be ripped apart by black-hat later on. One nice example that applies this quite nicely is roborev [0] by the legendary Wes McKinney. 0: https://github.com/roborev-dev/roborev