Reproduction Package - Automated Quality Assessment of User Stories Using Large Language Models: A Cautionary Study of Agreement Metrics on a Violation-Sparse Baseline

This deposit is the complete artifact: the Java evaluation pipeline (LangChain4j) under `pipeline/` and the exact prompt templates under `prompts/`, together with the dataset, the human baseline, the raw per-story model outputs, and the analysis scripts that reproduce every table and figure in the paper.