Spec2chat: A Python library for task-oriented dialogue generation from OpenAPI specifications

Abstract

This article presents spec2chat, a self-contained and pip-installable Python library for the automatic generation and execution of task-oriented dialogue systems from OpenAPI specifications enriched with conversational metadata. Unlike traditional frameworks based on handcrafted rules or LLM prompting, spec2chat enables dynamic dialogue generation directly driven by service definitions. It combines structured specifications with LLM-powered modules for natural language understanding and question generation. Its API-driven architecture supports hot-swapping of services and seamless adaptation to evolving APIs without retraining or manual intervention. Spec2chat provides a robust foundation for reproducible research and practical deployment in domains such as tourism, marketplaces, and digital public services.