This is a diagnostic framework for catching yourself when your reasoning is serving the wrong goal. It separates cognitive operations (differentiate, decouple, match, etc.) from what those operations serve, arguing that most thinking failures happen when good operations run in service of conclusion-preservation or threat-reduction instead of actual inquiry. Use it when you notice yourself explaining away evidence, when confidence feels defensive rather than earned, or when analysis keeps getting more elaborate without getting more accurate. The monitor/interrupt operation is the key mechanism, but there's a real trap here: under captured orientations, your self-correction machinery actively defends the wrong goal instead of fixing it. Heavy on metacognition, light on hand-holding.
npx skills add https://github.com/jwynia/agent-skills --skill good-thinking