contracts/real_data_pilot_intake_receipt.v2.schema.json

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://mparanza.com/schemas/clara/real-data-pilot-intake-receipt-v2.json",
  "title": "Clara real-data pilot intake receipt v2",
  "description": "A sanitized mechanical receipt showing that one exact trial-balance or general-ledger file matched a reviewed pilot declaration under its declared source kind. It is not proof that the declared kind is correct, legal authority, anonymization, semantic correctness, report readiness, or publication approval.",
  "$comment": "This schema validates portable structure and every expressible constant. validate_real_data_pilot_intake_receipt_v2 is authoritative for cross-field digest equality, date ordering, current implementation/dependency hashes, and other runtime invariants.",
  "type": "object",
  "additionalProperties": false,
  "required": [
    "schema_version",
    "pilot_id",
    "validation_date",
    "intake_contract_sha256",
    "source_receipt",
    "authorization_receipt",
    "privacy_boundary",
    "deidentification_review",
    "semantic_review",
    "eligibility",
    "validator"
  ],
  "properties": {
    "schema_version": {
      "const": "clara.real_data_pilot_intake_receipt.v2"
    },
    "pilot_id": {
      "$ref": "#/$defs/pilot_id"
    },
    "validation_date": {
      "type": "string",
      "format": "date"
    },
    "intake_contract_sha256": {
      "$ref": "#/$defs/sha256"
    },
    "source_receipt": {
      "$ref": "#/$defs/source_receipt"
    },
    "authorization_receipt": {
      "$ref": "#/$defs/authorization_receipt"
    },
    "privacy_boundary": {
      "$ref": "#/$defs/privacy_boundary"
    },
    "deidentification_review": {
      "$ref": "#/$defs/deidentification_review"
    },
    "semantic_review": {
      "$ref": "#/$defs/semantic_review"
    },
    "eligibility": {
      "$ref": "#/$defs/eligibility"
    },
    "validator": {
      "$ref": "#/$defs/validator"
    }
  },
  "allOf": [
    {
      "if": {
        "required": [
          "source_receipt"
        ],
        "properties": {
          "source_receipt": {
            "required": [
              "declared_data_classification"
            ],
            "properties": {
              "declared_data_classification": {
                "const": "anonymized_real"
              }
            }
          }
        }
      },
      "then": {
        "properties": {
          "deidentification_review": {
            "required": [
              "status",
              "reidentification_risk_review_status"
            ],
            "properties": {
              "status": {
                "const": "reviewed"
              },
              "reidentification_risk_review_status": {
                "const": "reviewed"
              }
            }
          }
        }
      }
    }
  ],
  "$defs": {
    "pilot_id": {
      "type": "string",
      "pattern": "^pilot-[0-9a-f]{16}$(?![\\s\\S])"
    },
    "source_id": {
      "type": "string",
      "pattern": "^source-[0-9a-f]{16}$(?![\\s\\S])"
    },
    "sha256": {
      "type": "string",
      "pattern": "^[0-9a-f]{64}$(?![\\s\\S])"
    },
    "source_receipt": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "source_id",
        "declared_data_kind",
        "declared_data_classification",
        "media_type",
        "byte_count",
        "sha256"
      ],
      "properties": {
        "source_id": {
          "$ref": "#/$defs/source_id"
        },
        "declared_data_kind": {
          "enum": [
            "commercial_general_ledger",
            "commercial_trial_balance"
          ]
        },
        "declared_data_classification": {
          "enum": [
            "anonymized_real",
            "consented_real"
          ]
        },
        "media_type": {
          "enum": [
            "application/vnd.ms-excel",
            "application/vnd.oasis.opendocument.spreadsheet",
            "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
            "text/csv"
          ]
        },
        "byte_count": {
          "type": "integer",
          "minimum": 1
        },
        "sha256": {
          "$ref": "#/$defs/sha256"
        }
      }
    },
    "authorization_receipt": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "status",
        "basis",
        "authorized_on",
        "valid_from",
        "valid_until",
        "purpose",
        "authorized_source_sha256",
        "permitted_actions",
        "prohibited_actions",
        "authorization_content_sha256",
        "assurance"
      ],
      "properties": {
        "status": {
          "const": "declared_reviewed"
        },
        "basis": {
          "enum": [
            "explicit_authorized_user_instruction",
            "public_owner_license",
            "written_data_owner_permission"
          ]
        },
        "authorized_on": {
          "type": "string",
          "format": "date"
        },
        "valid_from": {
          "type": "string",
          "format": "date"
        },
        "valid_until": {
          "oneOf": [
            {
              "type": "string",
              "format": "date"
            },
            {
              "type": "null"
            }
          ]
        },
        "purpose": {
          "const": "local_due_diligence_preparation_evaluation"
        },
        "authorized_source_sha256": {
          "$ref": "#/$defs/sha256"
        },
        "permitted_actions": {
          "const": [
            "codex_model_processing",
            "local_deterministic_processing"
          ]
        },
        "prohibited_actions": {
          "const": [
            "commit_raw_or_row_level_data",
            "package_raw_or_row_level_data",
            "publish_raw_or_row_level_data"
          ]
        },
        "authorization_content_sha256": {
          "$ref": "#/$defs/sha256"
        },
        "assurance": {
          "const": "declaration_bound_to_exact_source_not_independently_verified"
        }
      }
    },
    "privacy_boundary": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "codex_context_acknowledged",
        "automatic_anonymization_claimed",
        "clara_external_recipient_added",
        "declared_raw_and_row_level_storage",
        "repository_recording_policy",
        "storage_location_check"
      ],
      "properties": {
        "codex_context_acknowledged": {
          "const": true
        },
        "automatic_anonymization_claimed": {
          "const": false
        },
        "clara_external_recipient_added": {
          "const": false
        },
        "declared_raw_and_row_level_storage": {
          "const": "local_run_root_only"
        },
        "repository_recording_policy": {
          "const": "sanitized_summary_and_receipts_only"
        },
        "storage_location_check": {
          "$ref": "#/$defs/storage_location_check"
        }
      }
    },
    "storage_location_check": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "status",
        "source_relation",
        "intake_relation",
        "resolved_path_containment_enforced",
        "git_ignore_required_inside_repository"
      ],
      "properties": {
        "status": {
          "const": "passed_at_validation"
        },
        "source_relation": {
          "enum": [
            "inside_repository_git_ignored",
            "outside_repository_within_run_root"
          ]
        },
        "intake_relation": {
          "enum": [
            "inside_repository_git_ignored",
            "outside_repository_within_run_root"
          ]
        },
        "resolved_path_containment_enforced": {
          "const": true
        },
        "git_ignore_required_inside_repository": {
          "const": true
        }
      }
    },
    "deidentification_review": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "status",
        "reidentification_risk_review_status",
        "review_content_sha256",
        "assurance",
        "basis_recorded"
      ],
      "properties": {
        "status": {
          "enum": [
            "not_applicable",
            "reviewed"
          ]
        },
        "reidentification_risk_review_status": {
          "enum": [
            "not_applicable",
            "reviewed"
          ]
        },
        "review_content_sha256": {
          "$ref": "#/$defs/sha256"
        },
        "assurance": {
          "const": "review_presence_only_not_anonymization_certification"
        },
        "basis_recorded": {
          "const": true
        }
      }
    },
    "semantic_review": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "status",
        "required_review_count",
        "automatic_mapping_allowed",
        "unresolved_blocking_issues_block_preparation"
      ],
      "properties": {
        "status": {
          "const": "not_assessed"
        },
        "required_review_count": {
          "const": 7
        },
        "automatic_mapping_allowed": {
          "const": false
        },
        "unresolved_blocking_issues_block_preparation": {
          "const": true
        }
      }
    },
    "eligibility": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "status",
        "purpose",
        "publication_status",
        "report_ready",
        "execution_revalidation_required",
        "does_not_establish"
      ],
      "properties": {
        "status": {
          "const": "declared_boundary_passed_for_local_pilot_intake"
        },
        "purpose": {
          "const": "local_due_diligence_preparation_evaluation"
        },
        "publication_status": {
          "const": "withheld"
        },
        "report_ready": {
          "const": false
        },
        "execution_revalidation_required": {
          "const": true
        },
        "does_not_establish": {
          "const": [
            "accounting_semantic_correctness",
            "anonymization_sufficiency",
            "authorizer_identity_or_authority",
            "commercial_origin_or_authenticity",
            "current_authorization_at_later_execution",
            "declared_data_kind_correctness",
            "downstream_compatibility",
            "future_storage_or_copy_behavior",
            "legal_permission",
            "publication_authorization",
            "receipt_authenticity_or_signer_identity",
            "retention_or_deletion",
            "report_readiness"
          ]
        }
      }
    },
    "validator": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "dependency_sha256",
        "validator_id",
        "validator_version",
        "implementation_sha256",
        "mode"
      ],
      "properties": {
        "dependency_sha256": {
          "type": "object",
          "additionalProperties": false,
          "required": [
            "intake_receipt_schema",
            "intake_schema",
            "preparation_contract_kernel"
          ],
          "properties": {
            "intake_receipt_schema": {
              "$ref": "#/$defs/sha256"
            },
            "intake_schema": {
              "$ref": "#/$defs/sha256"
            },
            "preparation_contract_kernel": {
              "$ref": "#/$defs/sha256"
            }
          }
        },
        "validator_id": {
          "const": "real_data_pilot_intake_validator.v2"
        },
        "validator_version": {
          "const": "2.0.0"
        },
        "implementation_sha256": {
          "$ref": "#/$defs/sha256"
        },
        "mode": {
          "const": "deterministic_mechanical"
        }
      }
    }
  }
}