system-ASK — the ASK practice taken as a whole, apex and ecology together — is itself a concrete instance of the bounded-generativity topology defined in docs/bounded-generativity.md. The primitive's worked instances (§Worked instances) run from isometric-cubes-ASK through mazeASK, asset-pipeline-ASK, studioLeoV, generative-AI workflows, and the ASK advisory / extended-cognition loop. This example takes the last of those one altitude up: the same structure that governs a single advisory loop also governs the standing architecture of the whole practice.

This is a worked-example explanation at method altitude — not a second primitive definition and not an operational authorization. The primitive is defined in docs/bounded-generativity.md; this maps how system-ASK realizes it.

The bounded-generativity roles map onto system-ASK as follows:

• source of intent / normative apex — human-ASK: supplies purpose, the governing standard, authority, closure, and recourse.
• grammar and aperture — the question, the source-of-intent frame, the governing standard, constraints, open decisions, and the delegated scope that frame a given engagement.
• bounded realization field — ecology-ASK: advisor surfaces, execution surfaces, repositories, critics, evidence, and domain contributors, producing candidate distinctions and artifacts.
• authorized judgment where present — scoped expert, executor, or delegated decisions made inside the granted scope.
• selection + closure — ASK accepts, rejects, revises, holds, or relays what becomes operative.
• artifact governance — review, merge, canonicalization, absorption, provenance, and closure.
• governed output — the merged repo artifact, canonical note, adopted decision, or governed record.

system-ASK realizes the bounded-generativity topology at the practice level: human-ASK supplies the source of intent and closure; ecology-ASK supplies the bounded realization field; selected outcomes become durable through governance. In the primitive's relation-chain form:

order from chaos >> bounded generativity >> selection + closure >> artifact governance >> governed artifact + governance record

The recurrence is the point. The same topology holds at three scopes without changing shape: within a single generative artifact (isometric-cubes-ASK, mazeASK, image-making), within a single cognition loop (advisory / extended-cognition), and across the whole practice (this example). Second-and-beyond instances are what disambiguate project-specific shape from the cross-project type; the practice-level instance is the broadest of them.

This is a descriptive structural classification of system-ASK's topology. It does not claim that every system-ASK operation is correctly bounded, that every judgment is right, or that the system proves its own legitimacy — only that the roles, boundaries, and direction of authority have the bounded-generativity shape. The structure secures answerability in principle; whether it is discharged well in any given case is a separate, non-self-certifying question.

The direction of authority is invariant: intent originates at the apex and is not synthesized upward from the realization field. human-ASK sits inside the whole system-ASK boundary while remaining outside the bounded-generativity subframe it governs — the apex governing the field, not a paradox.

Ownership follows the layer model:

• method-ASK owns the primitive and this cross-surface structural relation.
• control-surface owns the execution protocols that move work across the relation — routing, handoffs, critique, review, instantiation, relay, and closure.
• project repos own their domain instances.