appreciate that. i generally don't do that selective context building step. i instead have claude plan, then run that plan through multiple iterations where i and/or codex reviews the plan and gets it to where it should be. but so often little things slip in that are maddening.
to be fair, i am not doing a line by line code review. for that i also defer to multiple iterations, where i'm reviewing each finding, but my scope is limited to what's surfaced by the agent.
but it's hard to believe big tech is reviewing line by line when they've pushing out thousands of lines.
indeed - human societies are very complex haha. we have managed to predict how social media react to messages at a 80%+ accuracy, but still early days in making the simulation accurate across all contexts and all populations :)
What's my point? That's a phenomenal question. I think my point was to see if anyone else out there shared my existential state. Now, of course I knew someone did, but there's really nothing like a good ol' confirmation from the other.
fyi https://docs.enkitask.com/ is not loading the js bundle. i'm on my corporate laptop behind a vpn, however, so maybe my company doesn't want me being efficient and effective :)
reply