write 1 bad prompt feed it to planner that reads the entire project into context and spits out 3 pages of plan
automagically send the plan to coder that reads the entire project into context and spits out 20k bad lines of code
automagically send the PR to tester that reads the entire project into context and tells coder he's a n00b and sends 50 red herrings as an eff-you
coder fixes all the 50 red herrings and then send it back to tester
tester pulls the feature branch and needs to load the entire thing into context again. is still pissy too because it doesn't work, back to coder: this shit dont work change <a>-<z>
coder runs out of context so needs to compact, gets all the new tokens, fixes the sole bug by accident
So that is 6 times your bill right there.
But then we use ralphwiggum(n=50) so it's actually 300x your bill, because every step gets retried 50x
The thing that makes it 😱 is:
plannerthat reads the entire project into context and spits out 3 pages of plancoderthat reads the entire project into context and spits out 20k bad lines of codetesterthat reads the entire project into context and tellscoderhe's a n00b and sends 50 red herrings as an eff-youcoderfixes all the 50 red herrings and then send it back totestertesterpulls the feature branch and needs to load the entire thing into context again. is still pissy too because it doesn't work, back tocoder: this shit dont work change<a>-<z>coderruns out of context so needs to compact, gets all the new tokens, fixes the sole bug by accidentSo that is 6 times your bill right there.
But then we use
ralphwiggum(n=50)so it's actually 300x your bill, because every step gets retried 50x