This skill provides a comprehensive framework for refactoring monolithic AI agent codebases into scalable, maintainable systems using Domain-Driven Design (DDD) and Clean Architecture principles. It focuses on decomposing 'god objects' into distinct bounded contexts—such as task management, session lifecycle, and health monitoring—while leveraging a microkernel pattern for extensible plugin support. By establishing strict domain boundaries and event-driven communication, this skill ensures your Claude-powered applications remain testable and organized even as system complexity increases.
主な機能
01Event-Driven Communication: Implements loose coupling between modules through domain events and an integrated event bus.
0259,149 GitHub stars
03Microkernel Plugin System: Provides a kernel-based architecture for dynamic domain loading and core extension.
04Modular Domain Decomposition: Breaks down monolithic orchestrators into isolated, focused bounded contexts.
05Clean Architecture Layers: Enforces separation between domain entities, application use cases, and infrastructure.
06Strategic Refactoring Patterns: Includes a phased migration strategy for safely extracting logic from existing legacy files.
ユースケース
01Implementing complex business logic and state transitions within isolated, testable domain aggregates.
02Refactoring 1000+ line orchestrator files into modular, domain-specific services.
03Building an extensible AI agent platform that supports dynamic plugin registration.