Maria Gabriela Alves Zuppardo, Vinicius Cardoso Garcia · 2026 · arXiv
DOI: 10.48550/arXiv.2610.31280 · open-access PDF
Los pipelines LLM multi-agente a menudo se acreditan con ganancias en tareas de ingeniería de software, pero rara vez está claro si un beneficio proviene del contrato tipado entre agentes o de las indicaciones especializadas que la descomposición permite. Estudiamos este problema de atribución en Ingeniería de Requisitos, comparando tres arquitecturas para clasificar los requisitos como funcionales o no funcionales y luego priorizándolos con MoSCoW: una línea de base de una sola llamada; una variante de dos llamadas que reutiliza las indicaciones especializadas del tubo pero pasa un intermediario no validado; y un tubo con un tipo validado por Pydantic.
Machine translation (Meta M2M100); the English original follows.
Multi-agent LLM pipelines are often credited with gains in software engineering tasks, but it is rarely clear whether a gain stems from the typed contract between agents or from the specialized prompts that decomposition enables. We study this attribution problem in Requirements Engineering, comparing three architectures for classifying requirements as functional or non-functional and then prioritizing them with MoSCoW: a single-call baseline; a two-call variant that reuses the pipeline's specialized prompts but passes an unvalidated intermediate; and a pipeline with a Pydantic-validated typed
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.31280, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange