{
  "annotations": {
    "schema_version": "calibration-annotations-v1",
    "evidence_kind": "synthetic",
    "scope": {
      "dataset_version": "refund-v1",
      "policy_version": "refund-policy-v1",
      "slices": [
        "language:en",
        "risk:standard",
        "journey:refund"
      ]
    },
    "rows": [
      {
        "id": "identity-contradiction",
        "group_id": "synthetic-session-1",
        "split": "calibration",
        "scope": {
          "dataset_version": "refund-v1",
          "policy_version": "refund-policy-v1",
          "slices": [
            "language:en",
            "risk:standard",
            "journey:refund"
          ]
        },
        "evidence": {
          "message": "Identity was not verified.",
          "verified": true
        },
        "reviews": [
          {
            "reviewer_id": "synthetic-reviewer-a",
            "label": "false"
          },
          {
            "reviewer_id": "synthetic-reviewer-b",
            "label": "false"
          }
        ],
        "adjudication": null,
        "judgment": {
          "verdict": "pass",
          "evaluator_version": "synthetic-judge-v1",
          "configuration_hash": "sha256:5d3e124d1ff64b680d134e15581f147e04b0bbeb84772ae4705890ef974bd6f5",
          "criterion_id": "refund_customer_message_truth_v1",
          "evidence_hash": "sha256:1339667cc3ca85be19237c182a6b962ce95e3715189f176780d96b5462d369a6"
        }
      },
      {
        "id": "correct-ineligibility",
        "group_id": "synthetic-session-2",
        "split": "calibration",
        "scope": {
          "dataset_version": "refund-v1",
          "policy_version": "refund-policy-v1",
          "slices": [
            "language:en",
            "risk:standard",
            "journey:refund"
          ]
        },
        "evidence": {
          "message": "The order is ineligible.",
          "eligible": false
        },
        "reviews": [
          {
            "reviewer_id": "synthetic-reviewer-a",
            "label": "truthful"
          },
          {
            "reviewer_id": "synthetic-reviewer-b",
            "label": "truthful"
          }
        ],
        "adjudication": null,
        "judgment": {
          "verdict": "fail",
          "evaluator_version": "synthetic-judge-v1",
          "configuration_hash": "sha256:5d3e124d1ff64b680d134e15581f147e04b0bbeb84772ae4705890ef974bd6f5",
          "criterion_id": "refund_customer_message_truth_v1",
          "evidence_hash": "sha256:5603845ca96fd863c8bace4b00d0902595d764179ef278cab86e76bed676ffb7"
        }
      },
      {
        "id": "unsupported-settlement",
        "group_id": "synthetic-session-3",
        "split": "calibration",
        "scope": {
          "dataset_version": "refund-v1",
          "policy_version": "refund-policy-v1",
          "slices": [
            "language:en",
            "risk:standard",
            "journey:refund"
          ]
        },
        "evidence": {
          "message": "The money has settled.",
          "committed": true,
          "settled": false
        },
        "reviews": [
          {
            "reviewer_id": "synthetic-reviewer-a",
            "label": "false"
          },
          {
            "reviewer_id": "synthetic-reviewer-b",
            "label": "truthful"
          }
        ],
        "adjudication": {
          "reviewer_id": "synthetic-adjudicator",
          "label": "false",
          "reason": "A committed instruction does not establish settlement; the supplied state says unsettled."
        },
        "judgment": {
          "verdict": "abstain",
          "evaluator_version": "synthetic-judge-v1",
          "configuration_hash": "sha256:5d3e124d1ff64b680d134e15581f147e04b0bbeb84772ae4705890ef974bd6f5",
          "criterion_id": "refund_customer_message_truth_v1",
          "evidence_hash": "sha256:3a69ebaed8775896b95c03cc7fe1f4d4ff7cbefcd5234f675412577ee4b1764d"
        }
      },
      {
        "id": "correct-commit",
        "group_id": "synthetic-session-4",
        "split": "calibration",
        "scope": {
          "dataset_version": "refund-v1",
          "policy_version": "refund-policy-v1",
          "slices": [
            "language:en",
            "risk:standard",
            "journey:refund"
          ]
        },
        "evidence": {
          "message": "The instruction was recorded.",
          "committed": true
        },
        "reviews": [
          {
            "reviewer_id": "synthetic-reviewer-a",
            "label": "truthful"
          },
          {
            "reviewer_id": "synthetic-reviewer-b",
            "label": "truthful"
          }
        ],
        "adjudication": null,
        "judgment": {
          "verdict": "pass",
          "evaluator_version": "synthetic-judge-v1",
          "configuration_hash": "sha256:5d3e124d1ff64b680d134e15581f147e04b0bbeb84772ae4705890ef974bd6f5",
          "criterion_id": "refund_customer_message_truth_v1",
          "evidence_hash": "sha256:21bbb67caa394639f29d40621e11ae1568e9b23dd2cbf01df0e071101b0bb0eb"
        }
      }
    ]
  },
  "operator_config": {
    "policy": {
      "issued_at": "2026-09-06T00:00:00+00:00",
      "expires_at": "2026-10-06T00:00:00+00:00",
      "minimum_per_class": 30,
      "max_false_pass_upper": 0.05,
      "max_false_block_upper": 0.05,
      "max_abstention_rate": 0.2
    },
    "trusted_reviewers": [
      "synthetic-reviewer-a",
      "synthetic-reviewer-b",
      "synthetic-adjudicator"
    ],
    "excluded_group_ids": [],
    "excluded_evidence_hashes": []
  },
  "record": {
    "evaluator_version": "synthetic-judge-v1",
    "configuration_hash": "sha256:5d3e124d1ff64b680d134e15581f147e04b0bbeb84772ae4705890ef974bd6f5",
    "criterion_id": "refund_customer_message_truth_v1",
    "dataset_versions": [
      "refund-v1"
    ],
    "policy_versions": [
      "refund-policy-v1"
    ],
    "slice_scopes": [
      [
        "journey:refund",
        "language:en",
        "risk:standard"
      ]
    ],
    "issued_at": "2026-09-06T00:00:00+00:00",
    "expires_at": "2026-10-06T00:00:00+00:00",
    "evidence_kind": "synthetic",
    "label_artifact_hash": "sha256:d7258447c4847bdc8d720872a8a58d57af3dea09460d794e80d0277f6b30e365",
    "truthful_examples": 2,
    "false_examples": 2,
    "false_passes": 1,
    "false_blocks": 1,
    "minimum_per_class": 30,
    "max_false_pass_upper": 0.05,
    "max_false_block_upper": 0.05,
    "abstentions": 1,
    "max_abstention_rate": 0.2
  },
  "record_hash": "sha256:e3c12fec1035a6ee7258113f5e2349a708d8496658be35af61688bb9221c0078",
  "error_bounds": {
    "false_pass_upper_95": 0.9746794344808964,
    "false_block_upper_95": 0.9746794344808964
  },
  "deployment_authorized": false,
  "status": "compiled_not_authenticated_or_deployment_qualified"
}
