{
  "schemaVersion": 1,
  "disclosure": "Hand-authored synthetic judge scores, reference assignments, outputs and permissions. PASS_DEMO_POLICY is not deployment approval, model quality evidence, human-calibration evidence, or security assurance.",
  "contract": {
    "knowledgeSnapshot": "synthetic-service-docs-v1",
    "querySetVersion": "release-demo-v1",
    "judgeVersion": "simulated-judge-v1",
    "rubricVersion": "authored-quality-rubric-v1"
  },
  "policy": {
    "scope": "synthetic-demo-only",
    "minimumMeanGain": 5,
    "maximumCriticalDrop": 0,
    "minimumAcceptedRateDelta": 0,
    "calibration": {
      "thresholdCandidates": [
        50,
        70,
        90
      ],
      "falseAcceptCost": 2,
      "falseRejectCost": 1,
      "tieBreak": "lowest-threshold"
    }
  },
  "splitAudit": {
    "development": 2,
    "calibration": 8,
    "replay": 6,
    "holdoutRole": 2,
    "uniqueCaseIds": 18,
    "familyIdsDisjoint": true,
    "heldoutGeneralizationClaim": false,
    "note": "Split names demonstrate workflow routing only; all samples and rules were co-designed and are public. Family IDs cannot detect semantic leakage."
  },
  "calibration": {
    "sourceSplit": "calibration",
    "sampleCount": 8,
    "falseAcceptCost": 2,
    "falseRejectCost": 1,
    "tieBreak": "lowest-threshold",
    "candidates": [
      {
        "threshold": 50,
        "counts": {
          "trueAccept": 4,
          "falseAccept": 2,
          "trueReject": 2,
          "falseReject": 0
        },
        "weightedError": 4,
        "agreement": 0.75,
        "acceptPrecision": 0.666666666667,
        "acceptRecall": 1,
        "trueAcceptIds": [
          "cal-complete",
          "cal-exact",
          "cal-qualified",
          "cal-terse"
        ],
        "falseAcceptIds": [
          "cal-fluent-wrong",
          "cal-stale"
        ],
        "trueRejectIds": [
          "cal-invented",
          "cal-missing-condition"
        ],
        "falseRejectIds": []
      },
      {
        "threshold": 70,
        "counts": {
          "trueAccept": 3,
          "falseAccept": 1,
          "trueReject": 3,
          "falseReject": 1
        },
        "weightedError": 3,
        "agreement": 0.75,
        "acceptPrecision": 0.75,
        "acceptRecall": 0.75,
        "trueAcceptIds": [
          "cal-complete",
          "cal-exact",
          "cal-qualified"
        ],
        "falseAcceptIds": [
          "cal-fluent-wrong"
        ],
        "trueRejectIds": [
          "cal-invented",
          "cal-missing-condition",
          "cal-stale"
        ],
        "falseRejectIds": [
          "cal-terse"
        ]
      },
      {
        "threshold": 90,
        "counts": {
          "trueAccept": 1,
          "falseAccept": 0,
          "trueReject": 4,
          "falseReject": 3
        },
        "weightedError": 3,
        "agreement": 0.625,
        "acceptPrecision": 1,
        "acceptRecall": 0.25,
        "trueAcceptIds": [
          "cal-exact"
        ],
        "falseAcceptIds": [],
        "trueRejectIds": [
          "cal-fluent-wrong",
          "cal-invented",
          "cal-missing-condition",
          "cal-stale"
        ],
        "falseRejectIds": [
          "cal-complete",
          "cal-qualified",
          "cal-terse"
        ]
      }
    ],
    "chosenThreshold": 70,
    "chosenWeightedError": 3,
    "residualFalseAcceptIds": [
      "cal-fluent-wrong"
    ],
    "residualFalseRejectIds": [
      "cal-terse"
    ]
  },
  "candidates": [
    {
      "name": "clean-contrast",
      "baseline": "baseline",
      "decision": "PASS_DEMO_POLICY",
      "summary": {
        "count": 8,
        "baselineScoreTotal": 550,
        "candidateScoreTotal": 732,
        "baselineMean": 68.75,
        "candidateMean": 91.5,
        "meanGain": 22.75,
        "improved": 8,
        "tied": 0,
        "regressed": 0,
        "baselineAccepted": 4,
        "candidateAccepted": 8,
        "baselineAcceptedRate": 0.5,
        "candidateAcceptedRate": 1,
        "acceptedRateDelta": 0.5
      },
      "bySplit": [
        {
          "split": "replay",
          "count": 6,
          "baselineScoreTotal": 410,
          "candidateScoreTotal": 552,
          "baselineMean": 68.333333333333,
          "candidateMean": 92,
          "meanGain": 23.666666666667,
          "improved": 6,
          "tied": 0,
          "regressed": 0,
          "baselineAccepted": 3,
          "candidateAccepted": 6,
          "baselineAcceptedRate": 0.5,
          "candidateAcceptedRate": 1,
          "acceptedRateDelta": 0.5
        },
        {
          "split": "holdout-role",
          "count": 2,
          "baselineScoreTotal": 140,
          "candidateScoreTotal": 180,
          "baselineMean": 70,
          "candidateMean": 90,
          "meanGain": 20,
          "improved": 2,
          "tied": 0,
          "regressed": 0,
          "baselineAccepted": 1,
          "candidateAccepted": 2,
          "baselineAcceptedRate": 0.5,
          "candidateAcceptedRate": 1,
          "acceptedRateDelta": 0.5
        }
      ],
      "gates": [
        {
          "id": "aggregate-judge-uplift",
          "pass": true,
          "minimumMeanGain": 5
        },
        {
          "id": "paired-critical-regressions",
          "pass": true,
          "maximumDrop": 0,
          "violatingCaseIds": []
        },
        {
          "id": "calibrated-acceptance-rate",
          "pass": true,
          "minimumRateDelta": 0
        },
        {
          "id": "context-and-citation-acl",
          "pass": true,
          "violatingCaseIds": []
        },
        {
          "id": "required-abstention",
          "pass": true,
          "violatingCaseIds": []
        }
      ],
      "blockingGateIds": [],
      "paired": [
        {
          "caseId": "h-backup-window",
          "split": "holdout-role",
          "critical": false,
          "baselineScore": 60,
          "candidateScore": 90,
          "delta": 30,
          "baselineJudgeAccept": false,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "h-cache-boundary",
          "split": "holdout-role",
          "critical": true,
          "baselineScore": 80,
          "candidateScore": 90,
          "delta": 10,
          "baselineJudgeAccept": true,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-current-version",
          "split": "replay",
          "critical": false,
          "baselineScore": 50,
          "candidateScore": 90,
          "delta": 40,
          "baselineJudgeAccept": false,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-exact-code",
          "split": "replay",
          "critical": true,
          "baselineScore": 90,
          "candidateScore": 95,
          "delta": 5,
          "baselineJudgeAccept": true,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-no-answer",
          "split": "replay",
          "critical": true,
          "baselineScore": 90,
          "candidateScore": 95,
          "delta": 5,
          "baselineJudgeAccept": true,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-synonym",
          "split": "replay",
          "critical": false,
          "baselineScore": 50,
          "candidateScore": 90,
          "delta": 40,
          "baselineJudgeAccept": false,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-tenant-boundary",
          "split": "replay",
          "critical": true,
          "baselineScore": 90,
          "candidateScore": 92,
          "delta": 2,
          "baselineJudgeAccept": true,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-two-facts",
          "split": "replay",
          "critical": false,
          "baselineScore": 40,
          "candidateScore": 90,
          "delta": 50,
          "baselineJudgeAccept": false,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        }
      ]
    },
    {
      "name": "unsafe-uplift",
      "baseline": "baseline",
      "decision": "BLOCK",
      "summary": {
        "count": 8,
        "baselineScoreTotal": 550,
        "candidateScoreTotal": 703,
        "baselineMean": 68.75,
        "candidateMean": 87.875,
        "meanGain": 19.125,
        "improved": 6,
        "tied": 0,
        "regressed": 2,
        "baselineAccepted": 4,
        "candidateAccepted": 7,
        "baselineAcceptedRate": 0.5,
        "candidateAcceptedRate": 0.875,
        "acceptedRateDelta": 0.375
      },
      "bySplit": [
        {
          "split": "replay",
          "count": 6,
          "baselineScoreTotal": 410,
          "candidateScoreTotal": 528,
          "baselineMean": 68.333333333333,
          "candidateMean": 88,
          "meanGain": 19.666666666667,
          "improved": 4,
          "tied": 0,
          "regressed": 2,
          "baselineAccepted": 3,
          "candidateAccepted": 5,
          "baselineAcceptedRate": 0.5,
          "candidateAcceptedRate": 0.833333333333,
          "acceptedRateDelta": 0.333333333333
        },
        {
          "split": "holdout-role",
          "count": 2,
          "baselineScoreTotal": 140,
          "candidateScoreTotal": 175,
          "baselineMean": 70,
          "candidateMean": 87.5,
          "meanGain": 17.5,
          "improved": 2,
          "tied": 0,
          "regressed": 0,
          "baselineAccepted": 1,
          "candidateAccepted": 2,
          "baselineAcceptedRate": 0.5,
          "candidateAcceptedRate": 1,
          "acceptedRateDelta": 0.5
        }
      ],
      "gates": [
        {
          "id": "aggregate-judge-uplift",
          "pass": true,
          "minimumMeanGain": 5
        },
        {
          "id": "paired-critical-regressions",
          "pass": false,
          "maximumDrop": 0,
          "violatingCaseIds": [
            "r-exact-code",
            "r-no-answer"
          ]
        },
        {
          "id": "calibrated-acceptance-rate",
          "pass": true,
          "minimumRateDelta": 0
        },
        {
          "id": "context-and-citation-acl",
          "pass": false,
          "violatingCaseIds": [
            "r-tenant-boundary"
          ]
        },
        {
          "id": "required-abstention",
          "pass": false,
          "violatingCaseIds": [
            "r-no-answer"
          ]
        }
      ],
      "blockingGateIds": [
        "paired-critical-regressions",
        "context-and-citation-acl",
        "required-abstention"
      ],
      "paired": [
        {
          "caseId": "h-backup-window",
          "split": "holdout-role",
          "critical": false,
          "baselineScore": 60,
          "candidateScore": 90,
          "delta": 30,
          "baselineJudgeAccept": false,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "h-cache-boundary",
          "split": "holdout-role",
          "critical": true,
          "baselineScore": 80,
          "candidateScore": 85,
          "delta": 5,
          "baselineJudgeAccept": true,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-current-version",
          "split": "replay",
          "critical": false,
          "baselineScore": 50,
          "candidateScore": 95,
          "delta": 45,
          "baselineJudgeAccept": false,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-exact-code",
          "split": "replay",
          "critical": true,
          "baselineScore": 90,
          "candidateScore": 60,
          "delta": -30,
          "baselineJudgeAccept": true,
          "candidateJudgeAccept": false,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-no-answer",
          "split": "replay",
          "critical": true,
          "baselineScore": 90,
          "candidateScore": 85,
          "delta": -5,
          "baselineJudgeAccept": true,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": true,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-synonym",
          "split": "replay",
          "critical": false,
          "baselineScore": 50,
          "candidateScore": 95,
          "delta": 45,
          "baselineJudgeAccept": false,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-tenant-boundary",
          "split": "replay",
          "critical": true,
          "baselineScore": 90,
          "candidateScore": 98,
          "delta": 8,
          "baselineJudgeAccept": true,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [
              "tenant-b-private"
            ],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        },
        {
          "caseId": "r-two-facts",
          "split": "replay",
          "critical": false,
          "baselineScore": 40,
          "candidateScore": 95,
          "delta": 55,
          "baselineJudgeAccept": false,
          "candidateJudgeAccept": true,
          "baselineSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          },
          "candidateSignals": {
            "forbiddenContextDocumentIds": [],
            "forbiddenCitationDocumentIds": [],
            "requiredAbstentionViolation": false,
            "unexpectedAbstention": false
          }
        }
      ]
    }
  ]
}
