I don’t use loops, skills, fancy orchestration, plugins or anything like that. I raw dog the agent with its default harness, I read the code, and I’m shipping more than ever to production for real projects used by real people. Sometimes if it’s a low risk part of the application I might skip reading the code but only sometimes. I think most are either lying about the amount of value they are creating or just having fun while understanding they aren’t creating value but failing to distinguish that to their audience.
Your CEO should be strong.
Your CTO should be wise.
Your COO should be wicked, cunning, of mysterious origins, fluent in the dark arts, blurry in pictures,
A common dynamic I observe with AI: it feels most impressive when you don’t know much about the subject, don’t care or don’t have a clear idea of what the you want.
This applies across design, code, legal, and more. If I don’t know code very well, every piece of code it writes feels very impressive.
Once you know what something should feel or look like, it becomes almost impossible to guide AI there. And you definitely can’t one-shot it.