i'm thinking of building my own agent harness with a very specific focus - optimizing for flow.

llms are the perfect tool to create a cognitive layer that keeps you maximally engaged near the boundary of your current intellectual capability, removing the low level boring parts