architecture-refactoring-strategylisted

# Architecture Refactoring Strategy Bundled commands use `$PLUGIN_ROOT` (`$env:PLUGIN_ROOT` in PowerShell; same path suffix) for the plugin root. Set it once: use the host's plugin-root variable when defined (Claude Code: `PLUGIN_ROOT="$CLAUDE_PLUGIN_ROOT"`), otherwise the absolute path of this plugin's root directory. Use when improving the architecture of the application or library code needs staged implementation rather than a one-shot rewrite. Module boundaries, layers, dependency direction, APIs, domain/data ownership, extension points, and runtime coupling are the subject. Coding agents are tools, not the architecture being designed. ## Inputs Gather current architecture map, hot paths, structure metrics, runtime topology, ownership topology, target quality attribute, constraints, tests, production signals, failure history, ADRs, and architecture principles. For architecture-significant structural changes, require a baseline before code edits: dirty-tree state, observed boundary evidence, pre-refactor probe or explicit reason it was skipped, representative change/runtime scenario, and the focused skills used from the router. ## Patterns Use the smallest safe pattern: - characterization tests before moving behavior; - branch by abstraction behind stable interface; - strangler fig for extracting flows; - anti-corruption layer for old model or external API coupling; - facade/adapter to stabilize callers; - parallel run, shadow read, dual write for data migration ri