key coverage audit derived

{
  "input": {
    "body": {
      "key": [
        "customer_id"
      ],
      "datasets": [
        {
          "csv": "customer_id\nC-001\nC-002\nC-003\n",
          "name": "crm"
        },
        {
          "csv": "customer_id\nC-001\nC-002\n",
          "name": "billing"
        },
        {
          "csv": "customer_id\nC-001\nC-004\n",
          "name": "warehouse"
        }
      ]
    },
    "type": "http",
    "method": "POST",
    "bodyType": "json"
  },
  "output": {
    "type": "json",
    "example": {
      "limits": {
        "maxRows": 5000,
        "maxListed": 200,
        "maxColumns": 64,
        "maxCsvBytes": 262144,
        "maxDatasets": 10,
        "maxFindings": 500,
        "maxCategories": 1000,
        "maxKeyColumns": 4
      },
      "passed": false,
      "complete": true,
      "datasets": [
        {
          "name": "crm",
          "rows": 3,
          "distinct_keys": 3
        },
        {
          "name": "billing",
          "rows": 2,
          "distinct_keys": 2
        },
        {
          "name": "warehouse",
          "rows": 2,
          "distinct_keys": 2
        }
      ],
      "findings": [
        {
          "key": "C-003",
          "code": "KEY_IN_ONE_DATASET_ONLY",
          "detail": "C-003 appears only in crm.",
          "dataset": "crm"
        },
        {
          "key": "C-004",
          "code": "KEY_IN_ONE_DATASET_ONLY",
          "detail": "C-004 appears only in warehouse.",
          "dataset": "warehouse"
        },
        {
          "key": "C-002",
          "code": "KEY_MISSING_FROM_SOME_DATASETS",
          "detail": "C-002 is present in crm, billing and missing from warehouse."
        }
      ],
      "request_id": "example-request",
      "service_id": "key_coverage_audit",
      "assumptions": [
        "Keys are canonicalised exactly, so the same identifier formatted differently across systems is one key.",
        "A key present in three of five datasets is reported as such. A pairwise comparison cannot express that, which is why coverage is counted across all datasets at once."
      ],
      "limitations": [
        "Key presence across the datasets supplied. Whether the rows behind a shared key agree is dataset_reconciliation, not this.",
        "fully_covered is true only when every dataset was fully read and every key appeared in all of them."
      ],
      "observed_at": "2026-10-02T00:00:00.000Z",
      "unique_keys": [
        {
          "key": "C-003",
          "only_in": "crm"
        },
        {
          "key": "C-004",
          "only_in": "warehouse"
        }
      ],
      "not_examined": [],
      "partial_keys": [
        {
          "key": "C-002",
          "present_in": [
            "crm",
            "billing"
          ],
          "missing_from": [
            "warehouse"
          ]
        }
      ],
      "dataset_count": 3,
      "finding_count": 3,
      "fully_covered": false,
      "service_version": "1.0.0",
      "findings_by_code": {
        "KEY_IN_ONE_DATASET_ONLY": 2,
        "KEY_MISSING_FROM_SOME_DATASETS": 1
      },
      "in_every_dataset": 1,
      "partially_covered": 1,
      "coverage_histogram": {
        "1": 2,
        "2": 1,
        "3": 1
      },
      "distinct_key_count": 4,
      "in_one_dataset_only": 2
    }
  }
}