Agent-skills-eval: Benchmark tool for measuring agent skill improvements

A new open-source evaluation framework that tests whether adding specific skills to AI agents actually improves their outputs. Rather than guessing at agent effectiveness, makers can now measure concrete gains—useful for anyone building or tuning multi-tool agents.