Developed and integrated the GoldenSwag Evaluation Benchmarks into the Aleph-Alpha-Research/eval-framework repository, focusing on expanding logical reasoning evaluation for machine learning models. This work introduced new GoldenSwag and GoldenSwag IDK tasks, extending validation-set-based evaluation and enabling few-shot prompting on the same validation data. The implementation involved end-to-end changes, including Python-based development, comprehensive test coverage, and thorough documentation updates. Emphasizing data analysis and reproducibility, the feature provides concrete benchmarks for logical reasoning, supporting improved model selection and research throughput. Collaboration was maintained through descriptive commits and co-authorship, ensuring high code quality and alignment with research team requirements.