{
  "version": "context-evaluation-report-v1",
  "syntheticOnly": true,
  "model": "gpt-6-luna",
  "generatedAt": "2026-10-23T10:01:47.0288047+00:00",
  "fixtureSha256": "B16CBC88C96C75761D4015700DEB21150FA76F32C9422FB79F5BD19234DFDFA0",
  "resultsSha256": "CEE67BB0EAB5314BA5A53E855CACCABD3846964276498E49CF9163273624E521",
  "compilerGatePassed": true,
  "preliminaryLiveGatePassed": true,
  "broadQualityClaimsAllowed": false,
  "independentReviewCompleted": false,
  "repetitionsPerFinalPrompt": 1,
  "scope": "20 hand-authored synthetic scenarios; one final model sample per mode/case. Rubric review by implementation agent, not independent human validation. P95 is per scenario (one has two scripted turns), not a production SLA. Costs are list-rate estimates, not invoices. No population/neuroscience effect established.",
  "modes": [
    {
      "mode": "recent-only",
      "cases": 20,
      "providerErrors": 0,
      "retrievalTp": 0,
      "retrievalFp": 0,
      "retrievalFn": 7,
      "retrievalPrecision": null,
      "retrievalRecall": 0.0,
      "compilerViolations": 0,
      "reviewed": 20,
      "rubricPassRate": 1.0,
      "criticalViolations": 0,
      "clarificationTurns": 17,
      "p95CaseMs": 12630,
      "meanReplyCharacters": 109.1,
      "inputTokens": 83681,
      "outputTokens": 7025,
      "estimatedUsd": 0.005437
    },
    {
      "mode": "summary",
      "cases": 20,
      "providerErrors": 0,
      "retrievalTp": 0,
      "retrievalFp": 0,
      "retrievalFn": 7,
      "retrievalPrecision": null,
      "retrievalRecall": 0.0,
      "compilerViolations": 0,
      "reviewed": 20,
      "rubricPassRate": 0.95,
      "criticalViolations": 0,
      "clarificationTurns": 16,
      "p95CaseMs": 14485,
      "meanReplyCharacters": 118.14,
      "inputTokens": 83930,
      "outputTokens": 7180,
      "estimatedUsd": 0.006142
    },
    {
      "mode": "v2",
      "cases": 20,
      "providerErrors": 0,
      "retrievalTp": 7,
      "retrievalFp": 0,
      "retrievalFn": 0,
      "retrievalPrecision": 1.0,
      "retrievalRecall": 1.0,
      "compilerViolations": 0,
      "reviewed": 20,
      "rubricPassRate": 0.95,
      "criticalViolations": 0,
      "clarificationTurns": 14,
      "p95CaseMs": 14463,
      "meanReplyCharacters": 110.76,
      "inputTokens": 89917,
      "outputTokens": 8393,
      "estimatedUsd": 0.007046
    }
  ]
}