exploration-judge

Evaluating exploration sub-findings for grounding and consistency involves judging each finding across multiple dimensions, producing a structured YAML verdict that guides downstream synthesis weighting and surfaces contested areas when disagreements arise.

Weighting exploration findings for synthesis requires producing structured verdicts that score each sub-finding on coverage, relevance, and actionability, enabling automated ranking and evidence-based decision-making by the synthesis layer.

Highlighting contested findings during exploration evaluation involves identifying disagreements across sub-findings, surfacing genuine complexity in the output verdict YAML so the synthesizer can account for conflicting evidence in the final decision.

Structured verdicts rank sub-findings without writing code by evaluating all sub-findings across multiple dimensions and outputting a self-contained, evidence-based YAML file that enables automated ranking and decision-making by the synthesis layer.

Evaluating exploration coverage and relevance requires judging sub-findings across five core dimensions: grounding, consistency, coverage, relevance, and actionability, ensuring a comprehensive assessment for synthesis weighting.