Implementation:Dagster io Dagster Asset Check Decorator

Knowledge Sources	Dagster Dagster Docs
Domains	Data_Engineering, Data_Quality
Last Updated	2026-02-10 00:00 GMT

Overview

Concrete decorator for defining asset quality checks provided by the Dagster core library.

Description

The @asset_check decorator transforms a Python function into an AssetChecksDefinition object that validates the contents of a target asset. The decorated function must return an AssetCheckResult indicating pass/fail status, optional metadata, and a severity level. When blocking=True, a failed check prevents downstream assets from materializing.

The decorator supports resource injection (via required_resource_keys or typed parameters), additional asset inputs for cross-asset validation, and partitioned checks that align with the target asset's partitioning scheme.

Usage

Import the decorator and result classes from the dagster package. Apply the decorator to a function that performs validation logic and returns an AssetCheckResult. Register the check in the Definitions object alongside assets.

Code Reference

Source Location

• Repository: dagster
• File: python_modules/dagster/dagster/_core/definitions/decorators/asset_check_decorator.py:L102

Signature

def asset_check(
    *,
    asset: Union[CoercibleToAssetKey, AssetsDefinition, SourceAsset],
    name: Optional[str] = None,
    description: Optional[str] = None,
    blocking: bool = False,
    additional_ins: Optional[Mapping[str, AssetIn]] = None,
    additional_deps: Optional[Iterable[CoercibleToAssetDep]] = None,
    required_resource_keys: Optional[set[str]] = None,
    metadata: Optional[Mapping[str, Any]] = None,
    automation_condition: Optional[AutomationCondition[AssetCheckKey]] = None,
    partitions_def: Optional[PartitionsDefinition] = None,
) -> Callable[[AssetCheckFunction], AssetChecksDefinition]

Import

from dagster import asset_check, AssetCheckResult, AssetCheckSeverity

I/O Contract

Inputs

Name	Type	Required	Description
asset	Union[CoercibleToAssetKey, AssetsDefinition, SourceAsset]	Yes	The target asset that this check validates.
name	str	No	The name of the check. Defaults to the function name.
description	str	No	Human-readable description displayed in the Dagster UI.
blocking	bool	No	If True, a failed check prevents downstream assets from materializing. Defaults to False.
additional_ins	Mapping[str, AssetIn]	No	Extra asset inputs for cross-asset validation.
additional_deps	Iterable[CoercibleToAssetDep]	No	Additional asset dependencies that do not pass data.
required_resource_keys	set[str]	No	Set of resource keys required by the check function.
metadata	Mapping[str, Any]	No	Static metadata attached to the check definition.
automation_condition	AutomationCondition[AssetCheckKey]	No	Condition under which the check should be automatically executed.
partitions_def	PartitionsDefinition	No	Partitioning scheme for the check, typically matching the target asset.

Outputs

Name	Type	Description
return	AssetChecksDefinition	A check definition object that can be included in a Dagster Definitions object.
(function return)	AssetCheckResult	The decorated function must return an AssetCheckResult with passed (bool), metadata (Mapping), and severity (AssetCheckSeverity).

Usage Examples

Null Value Check

import dagster as dg

@dg.asset_check(asset=raw_customers, blocking=True)
def no_null_customer_ids(duckdb: DuckDBResource) -> dg.AssetCheckResult:
    with duckdb.get_connection() as conn:
        null_count = conn.execute(
            "SELECT COUNT(*) FROM raw_customers WHERE customer_id IS NULL"
        ).fetchone()[0]
    return dg.AssetCheckResult(
        passed=null_count == 0,
        metadata={"null_count": null_count},
        severity=dg.AssetCheckSeverity.ERROR,
    )

Row Count Threshold Check

import dagster as dg

@dg.asset_check(asset=daily_orders, blocking=False, description="Warn if fewer than 100 orders")
def minimum_order_count(duckdb: DuckDBResource) -> dg.AssetCheckResult:
    with duckdb.get_connection() as conn:
        count = conn.execute("SELECT COUNT(*) FROM daily_orders").fetchone()[0]
    return dg.AssetCheckResult(
        passed=count >= 100,
        metadata={"order_count": count},
        severity=dg.AssetCheckSeverity.WARN,
    )

Related Pages

Implements Principle

• Principle:Dagster_io_Dagster_Data_Quality_Validation