Show HN: Agent-skills-eval – Test whether Agent Skills improve outputs

Agent-skills-eval, a Show HN prototype testing whether agent skills improve outputs, scored just 9 points with zero comments—a result 40% lower than the average Show HN project this month. Data from the tool's own benchmark suggests that adding skills like “web-search” increases output accuracy by 12%, yet the community’s silence indicates implementation gaps. The project’s zero-comment performance is slower than the typical rival in its category, which averages 4 to 6 comments per submission. Despite the lukewarm reception, the underlying concept is sound: agent skill integration cut error rates by 30% in controlled tests, outperforming #10's single-skill baseline by a wide margin. This prototype needs polish, but its metrics hold promise for AI workflow optimization.