Check reference¶

CatalystData ships 51 check types, generated from the code on every build - this page can't drift from what's actually available.

Table-level¶

Check	What it catches
Column count between (min/max columns)	Validates that a table's column count falls within expected bounds - catches accidental schema changes.
Column name exists (schema validation)	Validates that a table has a column with a given name.
Row count (30-day average or min/max)	Validates that a table's row count is within expected bounds, either fixed limits or a 30-day rolling average.

Check	What it catches
Anomaly: spike/drop in row count or column sum	Detects anomalies by comparing the current row count or column sum against recent history.
Composite unique	Validates that combinations of columns are unique, e.g. (tenant_id, email).
Cross-field math	Validates that a numeric column equals a formula computed over other columns in the same row.
Cross-table reconciliation (totals match)	Validates that an aggregate over a column matches the same aggregate in another table.
Custom SQL check (query + threshold)	Runs a custom SQL query and compares the result against a threshold - the escape hatch for anything not covered by a built-in check.
Null check	Validates that a column's null percentage stays within a threshold.
Referential integrity (FK exists in parent)	Validates that values in a column exist in a parent table's column.
Unique check	Validates that a column's duplicate percentage stays within a threshold.
Value in allowed options	Validates that every non-null value in a column is in a fixed set of allowed values.

Check	What it catches
Default value	Validates that non-null values in a column match an expected default value.
No NaN/Inf (float)	Validates that a float or double column has no NaN or infinite values.
Non-negative	Validates that every non-null value in a numeric column is zero or greater.
Outlier (IQR)	Flags statistical outliers using the interquartile range (IQR) method.
Outlier (z-score)	Flags statistical outliers using the z-score method.
Precision/scale (decimal)	Validates that decimal values fit within an expected precision and scale.
Range (min/max)	Validates that a numeric column's values fall within expected bounds.
Sign consistency	Validates that every value in a numeric column has the expected sign.
Skewness	Validates that a numeric column's sample skewness stays within an expected range.

Check	What it catches
Case (UPPER/lower)	Validates that non-null string values match an expected case, upper or lower.
Format / pattern (regex or LIKE)	Validates that non-null string values match a pattern, expressed as a regex or a SQL LIKE pattern.
Length (min/max, whitespace)	Validates string length bounds and optionally flags leading or trailing whitespace.
No truncation (max length)	Validates that no value is at or beyond the column's maximum length, catching possible truncation.
Not empty string	Validates that non-null string values aren't empty strings.
Not placeholder (N/A, UNKNOWN, etc.)	Validates that non-null values aren't placeholder text such as N/A or UNKNOWN.
Reference table (allowed categories)	Validates that non-null values exist in a reference table's column - a foreign key for string categories.

Check	What it catches
Date within range (min/max)	Validates that a date or datetime column's values fall within an expected range.
Freshness (e.g. updated within N hours)	Validates that a date or datetime column has recent values, catching data that has gone stale.
No future dates	Validates that a date or datetime column has no dates after now.
Start ≤ end (two columns)	Validates that one date column is never later than another, e.g. start_date before end_date.
Valid date/time (no Feb 30, etc.)	Validates that a date or datetime column contains only genuinely valid dates.

Check	What it catches
Boolean implies other field	Validates that a required column is non-null whenever a boolean column has a given value.
Only TRUE/FALSE or 1/0	Validates that a boolean column contains only recognized values.

Check	What it catches
Checksum/hash validation	Validates that a stored checksum or hash matches the value computed from the binary data.
Not NULL and not empty/corrupted	Fails when a non-null binary value has zero length.
Size (min/max, no zero-byte)	Validates binary column size against optional minimum and maximum byte bounds.
Valid binary format (magic number)	Validates that binary values start with an expected magic-byte signature, e.g. a file format header.

Check	What it catches
Document not empty	Validates that a JSON or XML column's non-null values aren't empty.
Required nodes/keys present	Validates that required keys or nodes are present in every document.
Schema validation (JSON Schema / XSD)	Validates documents against a JSON Schema or XSD.
Well-formed JSON/XML (no syntax errors)	Validates that a JSON or XML column's non-null values are syntactically well-formed.

Check	What it catches
No placeholder UUIDs (e.g. all zeros)	Validates that a UUID column has no placeholder values, e.g. all-zeros.
Valid UUID format (v1/v4)	Validates that non-null values are well-formed UUIDs.

Check	What it catches
Coordinates within valid ranges	Validates that geometry coordinates fall within valid longitude and latitude ranges.
Geometry well-formed (no self-intersecting polygons)	Validates that geometries are well-formed, catching things like self-intersecting polygons.
No NULL geometries	Validates that a geometry or geography column has no null values.
Within expected geographic boundaries	Validates that geometries fall within an expected longitude/latitude bounding box.

Check	What it catches
Amount and currency match region/account	Validates that a currency code column only contains an allowed set of codes.
Range (no negative unless allowed, no extreme values)	Validates that a currency amount column stays within bounds, and is non-negative unless explicitly allowed.