Function evaluateComparatorGate

src/eval/discovery-harness.ts:369–401

(
  summary: DiscoverySummaryType,
  comparatorName: string,
  protocol: DiscoveryBenchmarkProtocol,
  comparatorEvidence: DiscoveryComparatorEvidence | undefined
)

}

function evaluateComparatorGate(
  summary: DiscoverySummaryType,
  comparatorName: string,
  protocol: DiscoveryBenchmarkProtocol,
  comparatorEvidence: DiscoveryComparatorEvidence | undefined
): DiscoveryComparatorGateResult {
  const tolerancePercent = protocol.shipGate.comparators.tolerancePercent;
  const comparatorMetrics = comparatorEvidence?.[comparatorName];
  const comparisons = protocol.shipGate.comparators.usefulnessMetrics.map((metric) =>
    compareMetricWithinTolerance(
      getDiscoveryMetricValue(summary, metric),
      getDiscoveryMetricValue(comparatorMetrics, metric),
      metric,
      tolerancePercent
    )
  );
  const missingMetrics = comparisons
    .filter((comparison) => comparison.comparatorValue === null)
    .map((comparison) => comparison.metric);

  return {
    comparatorName,
    status:
      missingMetrics.length > 0
        ? 'pending_evidence'
        : comparisons.every((comparison) => comparison.passes)
          ? 'passed'
          : 'failed',
    tolerancePercent,
    missingMetrics,
    comparisons
  };
}

export function combineDiscoverySummaries(summaries: DiscoverySummaryType[]): DiscoverySummaryType {
  return summarizeDiscoveryResults(summaries.flatMap((summary) => summary.results));