A loop with a vague goal never ends — it just polishes forever. Learn to write a definition of done, set scope and non-goals, and build a reusable brief.
Welcome to Writing Goals a Loop Can Finish. In Module 1 you learned that every loop needs a place to stop. This module is about the thing that makes stopping possible: a goal the loop can actually finish. It sounds obvious, but it’s where most loops quietly fail. Ask an AI to “make the newsletter better” and it will happily keep making changes forever, because “better” has no finish line. The loop can’t stop because you never told it what done looks like.
Here you’ll learn to write goals that end. You’ll see why open-ended requests like “improve this” trap a loop in endless polishing. You’ll write a definition of done — success criteria specific enough that you, or the AI, can check them and get a clear yes or no. You’ll set scope and, just as importantly, state non-goals — the things the loop should deliberately leave alone. Then you’ll pull it all together into a brief: a short, reusable template that captures the goal, the done criteria, the constraints, and the boundaries for any loop you run.
By the end you’ll have converted five real, badly-worded Harborlight requests into loop-ready briefs — a skill you’ll lean on in every module that follows. Start with Lesson 1, where we look at why “make it better” never finishes.
Complete all 5 lessons to finish the Writing Goals a Loop Can Finish module.