# Health Checks

Health Checks provide a structured way to define and enforce **data quality expectations** directly within your transformation code. After each build, these expectations are automatically evaluated and presented in the **Health** tab, offering clear and immediate insight into the quality and reliability of your datasets.

### What are Health Checks?

A health check is a declarative rule that validates specific assumptions about your data, including but not limited to:

* Columns must not contain null values
* String values must not be empty
* Values must be unique
* Numeric values must fall within a defined range

Rather than implementing imperative validation logic, you explicitly declare the conditions that must hold true. DataSpace then evaluates these conditions automatically once the build has completed.

### How it works

Health checks are defined within a transformation using a declarative API. Checks are associated with individual columns and returned alongside the transformed dataset.

Once the build finishes:

* All defined checks are executed automatically
* Results are aggregated into an overall pass rate
* Detailed per-column results are displayed in the **Health** tab
* Warnings and failures are clearly indicated

For a complete list of supported checks and configuration options, refer to the API documentation.

### Severity levels

Each health check specifies a **severity level**, which determines its impact on the build outcome:

{% columns %}
{% column %}
{% hint style="warning" %}
**Warn**: The build completes successfully, but the issue is recorded and visible in the Health tab
{% endhint %}

{% endcolumn %}

{% column %}
{% hint style="danger" %}
**Fail**: The build is marked as failed, even if the transformation code itself executed successfully
{% endhint %}

{% endcolumn %}
{% endcolumns %}

Severity levels allow you to differentiate between non-blocking data quality issues and strict guarantees that must be satisfied for a build to be considered successful.