ResearchClawBench is a benchmark that measures whether AI coding agents can independently conduct scientific research — from reading raw data to producing publication-quality reports — and then ...