Principles

Overview

What lives in the principles registry

• The principles registry holds the framework's short declarative claims about how AI work goes well or badly. A principle, in the registry's sense, is a one-line claim about practice referenced from more than one canonical document and too useful to leave embedded in any single one. Patterns and aspects are the framework's structural vocabulary; principles are its load-bearing assertions about how practitioners ought to work, collected here as a single canonical reference rather than as repeated prose across the canon.
• The bar for canonisation is narrow. A statement earns a place when it is short enough to be stated in one sentence without losing force, is referenced from more than one canonical document, and is not already covered by an existing principle. The current registry groups entries by the discipline they bear on most heavily: strategic principles for forming an AI strategy, engineering principles for runtime governance, training and adoption principles for institutional change, and piloting principles for the bounded engagements that test pattern fit before scale.
• The registry is not closed. New principles are added when they prove cross-referenceable across the documentation; existing ones are sharpened when later canon reveals additional structure. Each entry carries a one-line statement, a short rationale, the patterns and disciplines it applies to most stringently, and the canonical sources that already reference it. The registry is a one-stop reference, not a substitute for the documents that elaborate each principle in context.