Reproducibility package for "Evaluation Criterion Sensitivity in Software Effort and Work Estimation: A Benchmark Across 26 Datasets with Leakage Auditing"

Version 1.1.1 reproducibility and evidence release for the manuscript "Evaluation Criterion Sensitivity in Software Effort and Work Estimation: A Benchmark Across 26 Datasets with Leakage Auditing." The archive contains processed analytical datasets and metadata for 26 software effort and work estimation datasets, saved out-of-sample predictions, optimizer traces, analysis code, report-reconstruction scripts, statistical outputs, figures, tests, environment files, protocol-lock information, the Use Case Points leakage audit, and SHA-256 checksums. It supports document verification, report reconstruction without estimator refitting, analysis replay from saved predictions, and complete model rerun under the documented environment. Third-party source datasets are not relicensed by this record. Where original source material is not redistributed, the archive provides provenance, public identifiers, hashes, processing information, and reacquisition instructions.