Capability-Oriented Software Engineering (COSE): An Incremental Methodology for Building Reusable Capability Libraries
Abstract
Software engineering projects frequently suffer from architectural re-invention, encapsulating functional knowledge within individual project silos. This paper presents Capability-Oriented Software Engineering (COSE), a formal incremental methodology designed to systematically convert software engineering effort into a cumulative, permanent library of reusable, decoupled capabilities. COSE establishes a mathematical framework for problem decomposition into single-responsibility micro-problems, resolved via deterministic or cognitive capabilities assembled into execution pipelines. Under COSE, large language models (LLMs) are treated strictly as interchangeable, bounded cognitive components within deterministic control structures, preserving model independence and auditability via persistent file-based artifacts. We evaluate COSE across a diverse spectrum of real-world production systems: Mammon (algorithmic trading), Knowledge Factory (hybrid cognitive distillation), Website Intelligence Factory (web reverse-engineering), an AI Middleware orchestrator, and commercial business applications (Dol-Goldur, Catalog Vision, and Andigraf). The empirical evidence demonstrates complete infrastructure decoupling, operational stability, and progressive reduction in marginal software construction costs