Your agent's test gate runs the tests in the tree the agent just edited

A post in this tag (Why AI Coding Agents Can't Grade Their Own Homework) proposes an out-of-context...