Reimagine — Advisory Target Architecture — refactor-chain · do-the-work (advisory)

Bundle: refactor-chain (self-diagnosing, self-healing fix-it pipeline). Phase: do-the-work (advisory branch) · Prerequisite: refactor-understand + (ideally) refactor-assessment-map · Next: nothing runs automatically — if and only if the user opts in, an approved step is handed to refactor-transform. Adaptivity / conditional: advisory-only and opt-in. This skill proposes; it never edits code. It is off the default path — the chain does behavior-preserving refactoring unless the human explicitly asks to reimagine.

Purpose

Behavior-preserving refactoring makes existing code cleaner without changing what it does or how it's shaped at the large scale. Sometimes that's not enough — the shape itself is the problem (a god-object, a tangled layering, a missing boundary). This skill steps back and proposes a bolder target design: two or three candidate architectures, honest trade-offs, a recommended one, and a sketch of how you'd migrate there incrementally. It is a thinking tool and a proposal document. It writes a proposal, not code. Nothing changes unless the human reads it, picks a direction, and explicitly opts in — at which point each approved step is executed by refactor-transform under the usual checkpoint→apply→verify safety loop.