Support treating dbt_utils.recency tests as first-class Freshness in Elementary OSS (including models)

[guillermojast]

Is your feature request related to a problem? Please describe.

Currently, Elementary OSS surfaces freshness primarily through:

• dbt source freshness (freshness: in sources.yml)
• Elementary’s own freshness_anomalies / event_freshness_anomalies tests

However, many teams use dbt_utils.recency as their standard freshness/SLA test — especially for models, since dbt source freshness only applies to sources.

Although dbt_utils.recency is conceptually a freshness test (and is classified as such in dbt Hub), in Elementary OSS it is treated as a generic test. As a result:

• Recency tests do not appear in the “Freshness” sections of reports/dashboards.
• Freshness coverage metrics may be incomplete when teams rely on recency.
• There is inconsistency between how dbt users think about freshness and how it is visualized in Elementary.

This creates confusion and fragmented observability when monitoring model freshness.

---

Describe the solution you'd like

Add native support in Elementary OSS to treat dbt_utils.recency tests as first-class freshness tests.

Possible implementation options:

1. Automatically classify tests using the dbt_utils.recency macro as freshness tests in Elementary metadata.
2. Alternatively (or additionally), allow classification via:
   • A specific test tag (e.g. tags: ['freshness'])
   • A configurable mapping in elementary.yml
3. Include these tests in:
   • Freshness-related report sections
   • Freshness coverage metrics
   • Any freshness dashboards or visualizations

Ideally, this would work for both sources and models, enabling a unified freshness monitoring approach across the project.

---

Describe alternatives you've considered

• Using elementary.freshness_anomalies instead of dbt_utils.recency.
  This works for anomaly-based freshness detection but does not replace explicit SLA-based freshness checks.

• Relying solely on dbt source freshness.
  This is insufficient because it does not support models.

• Forking/modifying Elementary OSS to reclassify recency tests manually.
  While technically possible, this creates maintenance overhead and diverges from upstream.

---

Additional context

Many teams standardize on dbt_utils.recency for freshness SLAs (e.g., “data must be updated within 2 hours”). Treating it as a first-class freshness test in Elementary would:

• Align Elementary with common dbt practices
• Improve consistency between test intent and observability presentation
• Enable model-level freshness visibility without relying exclusively on anomaly detection

This would make Elementary more flexible while maintaining backward compatibility.

---

Would you be willing to contribute this feature?

Yes — I would be open to contributing this feature with guidance on the appropriate place in the codebase to implement the test classification and reporting changes.