Sep 11, 2026
Great reframe! I was focusing on the runtime guardrails, and did not realize how much the discipline actually lives in the repo structure itself, before the agent comes into play.
One thing I need to play around with is creating starter projects that are AI ready. One feature I'd look at adding is include evals to prove what the AI can figure out on it's own. Then you'll know what the agent can't and put more guardrails and other engineering solutions in place as needed.
--
