evaluation-methodology

PluginEval scoring methodology evaluates skill quality across trigger accuracy, orchestration fitness, output quality, and scope calibration. It blends three evaluation layers using specific weights and applies Elo ranking to generate composite scores and badge thresholds.

Badge thresholds and dimension grades in PluginEval reports reflect rubric anchors and anti-pattern penalties. This Skill breaks down composite scores into plain-language meaning, explaining why a submission earns a specific badge tier like Bronze or Silver.

Anti-pattern penalties reduce scores for poor triggering or scope calibration, while Elo ranking dynamically positions skills against each other. This Skill explains how these factors blend within the evaluation layers to determine your final quality score.

To improve weak PluginEval dimensions, review the methodology's rubric anchors and address anti-pattern penalties. This Skill identifies whether your submission needs better triggering, tighter scope, clearer output structure, or stronger references to increase its score.

Yes, the evaluation methodology applies to reviewing trigger accuracy, orchestration fitness, output quality, and scope calibration for plugin and skill assessments. It provides actionable improvement guidance based on the specific rubric anchors and blend weights evaluated.