System Column Registry

Metaxy reserves a small set of system-managed columns that it attaches to feature metadata tables. These columns are part of the platform contract and are used by the metadata store, versioning engine, and migration tooling to keep track of feature lineage.

All system column names start with the metaxy_ prefix to avoid collisions with user-defined feature fields. Only the prefixed forms are supported.

Canonical column names

Canonical name	Explanation	Level	Type
metaxy_provenance_by_field	Derived from upstream data versions and code version per field	sample	struct
metaxy_provenance	Hash of metaxy_provenance_by_field	sample	string
metaxy_data_version_by_field	Defaults to metaxy_provenance_by_field, can be user-defined	sample	struct
metaxy_data_version	Hash of metaxy_data_version_by_field	sample	string
metaxy_feature_version	Derived from versions of relevant upstream fields	feature	string
metaxy_snapshot_version	Derived from the entire Metaxy feature graph	graph	string
metaxy_feature_spec_version	Derived from the part of the feature spec responsible for versioning	sample	string
metaxy_full_definition_version	Hash of the entire feature Pydanitc model schema and the Metaxy project	string	true
metaxy_created_at	Timestamp when the metadata row was created	sample	string
metaxy_materialization_id	External orchestration run ID (e.g., Dagster, Airflow) for tracking	run	string

Orchestration Integration

The metaxy_materialization_id column enables integration with external orchestration systems like Dagster, Airflow, or Prefect. This column allows you to track which orchestration run produced specific metadata records, simplifying debugging, monitoring, and filtering.