Inspect Data Meaning

Type: Function

Retrieve column-level governance information for a table. Returns data type classifications, sensitivity levels, PII flags, masking configurations, and applied policies for each column. Optionally filter to specific columns.

Use Cases

• An AI agent checks which columns contain PII before constructing a SELECT statement
• A security review audits masking rules applied to sensitive columns
• An agent determines which columns it can safely include in an output dataset
• A data engineer verifies that classification and masking are correctly configured after a policy deployment

Input Parameters

Parameter	Type	Required	Description
table_name	string	Yes	Fully qualified table name (DB.SCHEMA.TABLE).
columns	array	No	Array of column name strings to inspect. If omitted, all governed columns in the table are returned.

Output Schema

Returns an array of column objects:

Field	Type	Description
column_name	string	Name of the column
data_type	object	Semantic data type classification
data_type.id	string	Machine-readable type identifier (e.g., email_address, phone_number, full_name)
data_type.label	string	Human-readable label (e.g., "Email Address", "Phone Number")
sensitivity	string	Column sensitivity: public, internal, confidential, or restricted
is_pii	boolean	Whether this column is classified as personally identifiable information
masking	object	Masking configuration
masking.enabled	boolean	Whether masking is active on this column
masking.type	string	Masking type: hash, redact, partial, null, or none
masking.config	object	Additional masking parameters (e.g., {"visible_chars": 4} for partial masking)
masking.exempt_roles	array	Roles that bypass masking (e.g., ["DATA_ADMIN"])
policy	object	Policy that governs this column
policy.id	string	Policy identifier
policy.name	string	Human-readable policy name

Example Response

{
  "status": "success",
  "data": [
    {
      "column_name": "EMAIL",
      "data_type": {
        "id": "email_address",
        "label": "Email Address"
      },
      "sensitivity": "confidential",
      "is_pii": true,
      "masking": {
        "enabled": true,
        "type": "partial",
        "config": {
          "visible_chars": 4,
          "mask_char": "*",
          "preserve_domain": true
        },
        "exempt_roles": ["DATA_ADMIN", "COMPLIANCE_OFFICER"]
      },
      "policy": {
        "id": "pol_customer_pii_001",
        "name": "Customer PII Protection"
      }
    },
    {
      "column_name": "ACCOUNT_STATUS",
      "data_type": {
        "id": "categorical",
        "label": "Categorical"
      },
      "sensitivity": "internal",
      "is_pii": false,
      "masking": {
        "enabled": false,
        "type": "none",
        "config": {},
        "exempt_roles": []
      },
      "policy": {
        "id": "pol_customer_pii_001",
        "name": "Customer PII Protection"
      }
    }
  ],
  "errors": []
}

SQL Examples

Inspect all columns in a table

SELECT METATATE_APP.CORE.INSPECT_COLUMNS(
  OBJECT_CONSTRUCT('table_name', 'ANALYTICS_DB.CORE.CUSTOMERS')
);

Inspect specific columns

SELECT METATATE_APP.CORE.INSPECT_COLUMNS(
  OBJECT_CONSTRUCT(
    'table_name', 'ANALYTICS_DB.CORE.CUSTOMERS',
    'columns', ARRAY_CONSTRUCT('EMAIL', 'PHONE', 'SSN')
  )
);

Extract PII columns only

SELECT col.value:column_name::STRING AS column_name,
       col.value:sensitivity::STRING AS sensitivity,
       col.value:masking:type::STRING AS masking_type
FROM TABLE(FLATTEN(
  METATATE_APP.CORE.INSPECT_COLUMNS(
    OBJECT_CONSTRUCT('table_name', 'ANALYTICS_DB.CORE.CUSTOMERS')
  ):data
)) col
WHERE col.value:is_pii = TRUE;

JSON Request / Response (API)

Request:

{
  "method": "tools/call",
  "params": {
    "name": "inspect-data-meaning",
    "arguments": {
      "table_name": "ANALYTICS_DB.CORE.CUSTOMERS",
      "columns": ["EMAIL", "PHONE"]
    }
  }
}

Response:

{
  "content": [
    {
      "type": "text",
      "text": "{\"status\":\"success\",\"data\":[{\"column_name\":\"EMAIL\",\"data_type\":{\"id\":\"email_address\",\"label\":\"Email Address\"},\"sensitivity\":\"confidential\",\"is_pii\":true,\"masking\":{\"enabled\":true,\"type\":\"partial\",\"config\":{\"visible_chars\":4,\"mask_char\":\"*\",\"preserve_domain\":true},\"exempt_roles\":[\"DATA_ADMIN\"]},\"policy\":{\"id\":\"pol_customer_pii_001\",\"name\":\"Customer PII Protection\"}},{\"column_name\":\"PHONE\",\"data_type\":{\"id\":\"phone_number\",\"label\":\"Phone Number\"},\"sensitivity\":\"confidential\",\"is_pii\":true,\"masking\":{\"enabled\":true,\"type\":\"redact\",\"config\":{},\"exempt_roles\":[\"DATA_ADMIN\"]},\"policy\":{\"id\":\"pol_customer_pii_001\",\"name\":\"Customer PII Protection\"}}],\"errors\":[]}"
    }
  ]
}

Try it in the app

Open Metatate and navigate to the Test Tools tab to run inspect-data-meaning interactively. Enter a table name and optionally select specific columns to inspect.