harveyai/biglaw-bench ? reverse-engineered prompt

Build me a simple open source project for BigLaw Bench that helps people understand and run the benchmark for legal AI tasks.

I want a clean site and sample code that explains the three parts of the benchmark, Core, Workflows, and Retrieval, and lets someone browse example tasks and grading rubrics without getting lost. Make it easy to see what kinds of legal work are being tested, like drafting, research, due diligence, deal work, and document retrieval.

Please include clear pages or sections for the overview, sample datasets, and how scoring works, so it feels like a real benchmark people can explore. Use the existing repo structure as the starting point, and if you need anything current from docs or libraries, look it up online. I’m mostly looking for something polished, readable, and useful for someone who wants to evaluate a model on realistic legal work.