Definition
What is a verifiable stopping condition?
A verifiable stopping condition is a finish line a script or evaluator can check by reading real output: “the build exits with zero errors”, “every page in the list exists and contain valid schema”, “the checker marked every item PASS”. It is what lets an agent run unattended, because “done” becomes a fact instead of the agent’s opinion.
The failure mode it prevents is vagueness. It is the agent equivalent of a definition of done, which exists for the same reason: so finishing becomes a shared fact rather than one person’s judgment. “Improve our AI visibility” gives an agent no way to know when to stop, so it stops when its own output sounds finished, which is much earlier than actually finished. The discipline: if you cannot phrase the finish line as something checkable, the task is not ready to hand to an agent; narrow it until you can, and bind quality judgments to a separate checker’s PASS or FAIL.
Stub · This definition is being expanded with verified examples from shipped systems. Updated .
VibeMarketers publishes one verified marketing agent system every week, with the installable files.