Measure-before-you-ship discipline for your AI Code: a methodology skill, an experiment-reviewer agent, and a drop-in benchmark harness scaffold.
2 in the Claude listing