The initial findings from CodeScaleBench, a new benchmark designed to evaluate coding agents against the true complexity of enterprise software development, including large codebases and multi-repository tasks.

Full article content could not be extracted automatically. Read the original below.