For several months, I've been improving Spec Driven Development. When a plan is thrown in, sub-agents divide each part of the plan into features and then convert them into specs. Each spec is encapsulated as much as possible so that it can be assembled. Finally, it is checked with a checklist (created based on Todo when creating the spec, such as Test_(spec name).md) and returned to the user.
The problem was that the progress was slow and problems kept arising, requiring a lot of rework. Especially, due to the collapse of sub-agents and the addition of an open-close agent to monitor and warn about it, things eventually became unmanageable.
I couldn't handle it anymore, so two weeks ago I gave up and took a break. Just looking at the letter "A" of AI made me angry...so I stopped looking.
So I stayed away from AI. (Harmful..sob..)
At this point in writing, I have come up with a simpler idea for orchestration (I believe that human work should be minimized as much as possible). Based on this, I'm going to make a fresh start.
A diary should be written in a diary...ㅠㅠ
