Developed the MACHIAVELLI Benchmark Registry and accompanying documentation for the UKGovernmentBEIS/inspect_evals repository, focusing on standardizing the evaluation of agents’ ethical behavior in game-based scenarios. The work involved updating the metadata schema to include new fields, such as those supporting anthropic model flags, and refining evaluation workflows to improve reproducibility and governance metric assessment. Leveraging Python scripting, data analysis, and benchmarking skills, the developer generated and revised Markdown and YAML documentation to align with the new schema. These contributions enable more consistent, transparent, and scalable governance evaluations, supporting broader adoption of the MACHIAVELLI benchmark across research and evaluation teams.