Dagster 1.13.15 - Snowflake dbt Project Preview

Dagster 1.13.15, published on July 23, 2026, adds the preview SnowflakeDbtProjectComponent to dagster-snowflake for native orchestration of dbt projects on Snowflake. This is the main data engineering change in a small release that also adds documentation for the Dagster MCP server.

The full release notes and downloads are on the GitHub release page.

The new SnowflakeDbtProjectComponent gives Dagster a component dedicated to dbt projects running on Snowflake. The release places it in dagster-snowflake and describes the orchestration path as native. For teams that already use Dagster, dbt, and Snowflake together, this is the one addition worth isolating in a test deployment.

The scope in the notes is deliberately narrow. There are no listed configuration keys, command examples, migration steps, or compatibility constraints. The notes also do not deprecate an existing dbt integration. Operators should therefore treat the component as an additional integration path, not as a required replacement for current jobs or assets.

That distinction keeps upgrade work bounded. Existing pipelines have no documented migration task in this release. Teams evaluating the component can compare it against their current orchestration path without coupling that evaluation to the core upgrade.

The SnowflakeDbtProjectComponent is marked Preview, but Dagster 1.13.15 itself is not a prerelease. It is neither a release candidate nor a beta according to the release metadata. The stability label applies to the new component, not to the complete Dagster release.

The notes make no promise about the preview API, long term compatibility, or a schedule for general availability. That leaves production readiness with the operator. Test the exact paths your deployment uses, including definition loading, dbt project selection, Snowflake authentication, failure reporting, and retries. None of those behaviors are described by this release entry, so they should not be inferred from the word native.

Version tracking also needs both sides of the release name. Core ships as 1.13.15, while the libraries ship as 0.29.15. Record both versions in an evaluation result so a later component change can be tied to the package set that was actually tested.

The MCP entry is documentation only ¶

The second change adds documentation for the Dagster MCP server. The release notes do not claim a new MCP server feature, protocol change, configuration key, or runtime fix. This is a documentation update, and it should be read as such.

That matters when planning a rollout. An operator looking for an MCP behavior change will not find one documented here. The entry improves the path to existing information, but it does not provide evidence that a deployed Dagster process must be changed for MCP use.

Only two changes are listed: the Snowflake dbt preview component and the MCP documentation addition. There are no documented bug fixes, deprecations, breaking changes, schema migrations, or security fixes in the release notes. A normal dependency and smoke test pass is still appropriate, but the notes identify no special migration procedure.

For data platform teams, the decision is straightforward. Upgrade core and libraries through the usual validation path, then assess SnowflakeDbtProjectComponent separately if Snowflake hosted dbt orchestration is relevant. The preview label is the main constraint, and the short notes leave detailed behavior to direct testing.