Transform My Repo

Migrations die from optimism: the equivalent library that doesn't exist, the runtime behavior nobody knew was load-bearing, the big-bang rewrite that never ships. This skill plans transformations the evidence-first way — and its most valuable possible output is a well-argued "don't."

Division of labor: this skill answers whether to migrate, how hard, in what order, and with what safety net. The detailed implementation plan for each migration phase is a deep-plan run — do not duplicate that work here.

The Prime Directive (family rule)

No claim without evidence — on BOTH sides. Source facts need file:line. Target-stack facts (a library exists, a mechanism behaves a certain way) need a registry/docs citation with version, checked NOW — training-data memory of an ecosystem is not evidence, and "surely Python has an equivalent" is how migrations lose months.