Dagster 1.13 introduces several major improvements: an open-source AI skills repository (dagster-io/skills) for tools like Claude Code and Codex to better assist with Dagster development; partition-aware asset checks that let teams validate specific data partitions rather than whole datasets; virtual assets (experimental) for representing database views and other auto-updating logical assets; state-backed components enabled by default for integrations like dbt, Fivetran, and Airbyte; and 20+ new integration components covering Azure, GCP, Databricks, Tableau, and more. The dg CLI also received expanded API surface for programmatic and AI-agent use.
Table of contents
Better support for AI-assisted Dagster developmentPartitioned asset checks are here!A stronger developer experience with dgState-backed components are enabled by defaultPreview support for virtual assetsEasier adoption with 20+ additional componentsDagster+ improvementsLooking forward1 Impression