Dagster

New

• Added an owners parameter to define_asset_job, mirroring the field on regular jobs.
• Improved error messages when required config is missing in Dagster jobs, with clearer guidance on which fields are missing and example config to provide.
• LOGS_CAPTURED event messages no longer say "Started capturing logs", since compute log managers typically upload logs on completion rather than streaming in real time.
• [ui] Added client-side validation for the deployment settings YAML editor that warns when unknown keys are detected.
• [ui] Op tags are now shown on the asset details page in a separate "Op tags" row, alongside the asset's own tags.
• [ui] Redesigned the sensor dry run dialog with a stacked cursor layout, copy buttons for cursor values, and an inline run config preview.
• [dagster-clickhouse] Added new dagster-clickhouse, dagster-clickhouse-pandas, and dagster-clickhouse-polars libraries with native ClickHouse resources, IO managers, and dg components.
• [dagster-cloud] The EcsUserCodeLauncher now accepts a repository_credentials config option, allowing ECR credentials to be configured at the agent or deployment level instead of only per code location.
• [dagster-postgres] Added workload identity federation (WIF) authentication for PostgreSQL via a new auth_provider config option (azure_wif, gcp_wif, or aws_wif), with optional extras dagster-postgres[azure], dagster-postgres[gcp], and dagster-postgres[aws]. The Helm chart supports WIF via global.postgresqlAuthWifEnabled. (Thanks, @JohnMav!)

Bugfixes

• Fixed an error when starting a schedule whose name was previously used for a sensor (or vice versa).
• Fixed a race condition in Pipes that could cause dynamically-added log readers to be missed.
• [dg] Fixed bugs in the GitLab hybrid scaffolder used by dg plus deploy configure.
• [ui] Fixed the open/close behavior of the partition selector in the materialization dialog.
• [ui] Fixed false-positive syntax errors in asset selection inputs while the autocomplete dropdown was open.
• [ui] Fixed key:value autocompletion in search inputs not advancing into value-suggestion context when accepting a key with the Enter key.
• [ui] Fixed the asset catalog "Group by Owner" and "Group by Tags" views not showing assets with empty owner or tag arrays under a "None" section.
• [ui] Fixed asset selection summary tiles with long titles visually overlapping with adjacent UI.
• [ui] Fixed monospace text in several places that had reverted to the default font.
• [dagster-airbyte] Fixed errors parsing timestamps from newer versions of the Airbyte API.
• [dagster-cloud] Fixed an issue where the --defs-state-info argument was applied to the multipex server command but not to the gRPC server command in serverless deployments.
• [dagster-dbt] Fixed an issue where dbt assets configured with enable_duplicate_source_asset_keys could emit duplicate dependency entries and ambiguous source metadata when multiple dbt sources resolved to the same asset key.
• [dagster-k8s] Added a workaround for an upstream issue in the Kubernetes Python client where setting the NO_PROXY or no_proxy environment variable did not affect the configuration of Kubernetes API calls.
• [dagster-snowflake] Replaced deprecated Pydantic validator usage with field_validator to eliminate Pydantic deprecation warnings.

Documentation

• Added documentation for writing unit tests for declarative automation conditions.
• Added documentation for branch deployment RBAC.
• Added a dedicated documentation page for branch deployment concurrency settings.
• Documented the new asset-event webhook tokens.
• Clarified asset metric monitor behavior in the alerts documentation.
• Added required API scopes to the Databricks connection documentation.
• Updated the Databricks connection documentation with an OAuth flow for service principals.
• [dg] Documented the dg CLI configuration option for setting a custom virtual environment path.

The Dagster+ Terraform provider lets platform teams manage deployments, access controls, alerting, and more as code. Define entire environments declaratively, review changes through pull requests, and integrate Dagster+ into your existing infrastructure workflows.

New

• [ui] Users can now hold Cmd or Ctrl while clicking launch, materialize, re-execute, or retry buttons to open the resulting run in a new browser tab instead of navigating away from the current page.
• [ui] Added an Azure DevOps kind tag icon for assets tagged with the azuredevops kind.
• [dg] The dg utils integrations sub-command has been removed.
• [dagster-databricks] DatabricksClientResource now accepts a credentials_strategy argument, enabling federated and custom authentication flows via the Databricks SDK's CredentialsStrategy protocol. (Thanks, @hbellur0526!)

Bugfixes

• Fixed an issue where the asset daemon could fail when encountering foreign sensor cursors in automation condition sensor state.
• Fixed an issue where run retries were sometimes not executed by the event log consumer daemon.
• Fixed an issue where an invalid multi-partition key format could cause errors when computing partition statuses.
• [dg] Fixed an issue where the --api-token flag was ignored when passed to dg CLI commands.
• [dg] Fixed an issue in dg api asset-checks where the command used a non-existent top-level GraphQL resolver path.
• [ui] Fixed incorrect font rendering caused by a CSS variable name mismatch.
• [ui] Fixed an issue where the run config editor dialog displayed at an incorrect height.
• [ui] Fixed an issue where clicking asset selection autocomplete suggestions could cause the input to lose focus.
• [ui] Fixed an issue where the Users table in Org Settings did not support horizontal scrolling.
• [dagster-cloud-cli] Fixed GitHub Enterprise authentication for branch deployment CI/CD workflows.
• [dagster-databricks] Fixed a TypeError when credentials_strategy was None.
• [dagster-dbt] Fixed a thread-safety issue in concurrent metadata fetching that could cause intermittent errors.
• [dagster-snowflake-polars] Fixed an issue where dynamic partition keys were not properly escaped.

Dagster Plus

• [ui] Branch deployments now expose run queue settings in the run queue configuration dialog.

AI agents that only understand business definitions without knowing whether the underlying pipeline actually succeeded are confidently wrong and operational context from the orchestrator is the missing piece.

Once your pipelines span multiple Databricks workspaces, you're no longer orchestrating a single system you're coordinating a distributed one.

New

• Added PipesCompositeMessageReader (preview) to support multiple concurrent message streams in a single Pipes session.
• Added sensor:, schedule:, and job: attribute support to the asset selection syntax (e.g., sensor:my_sensor, job:my_job).
• Added automation_type: attribute support to the asset selection syntax, allowing queries like automation_type:schedule or automation_type:sensor. (Thanks, @bengotow!)
• State-backed integration components (e.g., AirbyteWorkspaceComponent, FivetranWorkspaceComponent) now default to LOCAL_FILESYSTEM state storage instead of legacy_code_server_snapshots.
• [dg] Added dg api issue create and dg api issue update commands.
• [dg] Added filter support to dg api issue list.
• [ui] Improved asset selection autocomplete performance.
• [dagster-dbt] DagsterDbtTranslatorSettings.enable_source_metadata now defaults to True, enabling upstream asset key remapping based on dbt source table names by default.

Bugfixes

• Fixed a SQL injection vulnerability in several IO managers. We recommend all users upgrade. More details can be found at https://github.com/dagster-io/dagster/security/advisories/GHSA-mjw2-v2hm-wj34.
• Fixed an issue where runs executed via execute_in_process() would sometimes fail to display the step timeline in the Dagster UI.
• Fixed a bug where multi-asset definitions containing virtual assets would produce incorrect execution plans.
• [ui] Fixed the tick result button for sensors using dynamic partitions.
• [dagster-aws] PickledObjectS3IOManager now defaults the S3 key prefix to an empty string when none is provided. (Thanks, @aksestok!)
• [dagster-databricks] PipesDatabricksClient.run_multi_task and PipesDatabricksServerlessClient.run_multi_task now give each submitted task its own message destination by default, fixing chunk-file collisions between concurrent tasks.
• [dagster-dbt] Fixed the dbt Cloud v2 polling sensor and adhoc job to use stable, ID-based identifiers, preventing naming conflicts.
• [dagster-dbt] Fixed the dbt Cloud v2 polling sensor to correctly filter out runs from adhoc jobs.

Documentation

• Added a guide for virtual assets.
• Added documentation for partitioned asset checks.
• Added documentation for the free_slots_after_run_end_seconds concurrency configuration option.
• Clarified SCIM provisioning and SSO permission requirements in the authentication documentation.
• Added a guide for migrating Dagster+ from the US to the EU control plane.

Major Changes Since 1.12.0

• AI-assisted development: Released dagster-io/skills, a collection of Dagster-focused AI skills for coding agents like Claude Code, OpenAI Codex, and others. Expanded dg api commands for programmatic inspection of assets, runs, jobs, schedules, and more.
• Partitioned asset checks: Asset checks can now target specific partitions of an upstream asset, aligning data quality logic with how partitioned data is produced and monitored.
• State-backed components enabled by default: Integrations that depend on external metadata (dbt, Fivetran, Airbyte, Tableau, Looker, etc.) now use persisted local state by default, providing a more predictable code location loading experience.
• Virtual assets (preview): New is_virtual parameter on @asset and AssetSpec for modeling assets like database views that automatically reflect upstream changes without explicit materialization.
• 20+ new components: Added or expanded components for dbt Cloud, Spark, Azure (Blob Storage, ADLS2), GCP (BigQuery, GCS, Dataproc), Databricks, Tableau, Looker, Census, Polytomic, and more. Integrations gained richer observability, metadata, and operational support.
• Deeper integration support: dbt Cloud supports partitioned assets; Databricks gained job-level subsetting and auto-cancel on run termination; Fivetran added polling sensors, retry-on-reschedule, and resync support; BI integrations auto-enrich assets with table metadata for cross-system lineage.
• Dagster+ improvements: Organization-level timezone settings, service users for Pro accounts, more resilient code server redeploy behavior, improved agent failure recovery, and expanded insights and alerting workflows.

Breaking Changes

• Removed deprecated external_asset_from_spec and external_assets_from_specs. Use AssetSpec inputs directly to Definitions(...) or AssetsDefinition(specs=[...]) instead.
• Removed deprecated single-AssetKey deps argument support from asset dependencies. Use a sequence of AssetDep objects instead.
• Removed deprecated get_all_asset_specs from Definitions.
• Removed deprecated legacy_freshness_policy parameter from @observable_source_asset.
• Removed deprecated auto_observe_interval_minutes parameter from @observable_source_asset.
• Removed deprecated legacy_freshness_policies_by_output_name parameter from AssetsDefinition.
• Removed deprecated load_component_at_path from ComponentLoadContext. Use context.load_component instead.
• Removed deprecated build_defs_at_path from ComponentLoadContext.
• [dagster-airbyte] Removed deprecated AirbyteState enum (use AirbyteJobStatusType instead) and removed deprecated legacy_freshness_policy and auto_materialize_policy parameters from build_airbyte_assets().
• [dagster-looker] Removed deprecated DagsterLookerResource.build_defs, get_asset_key, get_dashboard_asset_key, get_explore_asset_key, get_view_asset_key methods, and Type[DagsterLookerApiTranslator] support from API helpers.
• [dagster-powerbi] Removed deprecated PowerBIWorkspace.build_defs(), translator key helpers (use get_asset_spec() instead), and Type[DagsterPowerBITranslator] support in load_powerbi_asset_specs() (pass an instance instead).
• [dagster-sigma] Removed deprecated SigmaOrganization.build_defs(), DagsterSigmaTranslator.get_asset_key() (use get_asset_spec(...).key instead), and Type[DagsterSigmaTranslator] support in load_sigma_asset_specs() (pass an instance instead).

New

• (Preview) Added support for virtual assets. The @asset decorator and AssetSpec now accept an is_virtual parameter for defining assets that represent views or derived tables that don't need to be materialized. Virtual assets are supported in staleness calculations, execution planning, and declarative automation.
• Job-level config defaults are now applied when partial config is provided to a run.
• [dagster-dbt] Added enable_dbt_views_as_virtual_assets setting to DbtTranslatorSettings for automatically treating dbt views as virtual assets.

Bugfixes

• Fixed an issue where a sensor targeting a job with run_tags and specifying an asset_selection in the RunRequest would not apply the job's run_tags to the resulting run.
• Fixed a potential error in YAML config snapshot conversion when encountering None fields.
• [dg] Fixed dg plus deploy configure generating a GitHub Action that used Docker instead of the PEX build strategy.
• [dagster-cloud-cli] PR comments in CI are now scoped by deployment name, preventing overwrites across deployments.
• [ui] Fixed "Missing" partition selection for time-based partitioned assets.
• [ui] Fixed raw log display rendering after ansi-to-react library update.
• [ui] "Terminate all runs" dialog now handles extremely large sets of runs more reliably.

Documentation

• Added Dagster+ agent configuration page for serverless and hybrid deployments.
• Added data portability documentation page.
• Reorganized run isolation documentation for hybrid and serverless deployments.
• Added ELT pipeline example with dlt and Sling.

Dagster Plus

• Added percent-change comparison type for metric monitor alerts.
• Added SCIM Groups filter support for members.value eq queries.
• Added MCP server integration providing AI assistants with access to Dagster Cloud operations.
• Fixed an issue where the Dagster+ Kubernetes agent would emit log noise about DAGSTER_CLOUD_RAW_GIT_URL and DAGSTER_CLOUD_GIT_URL environment variables when onlyAllowUserDefinedK8sConfigFields was set.
• Fixed incorrect alert type label for metrics alerts.
• Fixed Databricks connections failing with an "unable to start warehouse" error.

Introduces Dagster skills, partitioned asset checks, state backed components, virtual assets, and stronger integrations.

How we configure Copybara for bi-directional syncing to enable a hub-and-spoke model for Git repositories

New

• [dg] dg projects can now configure agent_queue and image in pyproject.toml under [tool.dg.project], which are included in the generated dagster_cloud.yaml when running dg plus deploy.
• [dg] Added dg api job list and dg api job get commands for querying job metadata including schedules, sensors, and tags.
• [dg] Added dg api asset-check list, dg api asset-check get-executions, and dg api asset get-partition-status commands.
• [ui] The asset overview page now shows a loading spinner for "Latest materialization" while data is loading, instead of a dash.
• [ui] Removed the legacy asset catalog component and the flagAssetCatalogSidebar feature flag. The asset catalog sidebar is now always enabled.
• [ui] The pipeline snapshot link is no longer shown in the Run Actions Menu for hidden asset group jobs.
• [dagster-aws] The EcsRunLauncher and Dagster+ ECS agent now supply idempotency tokens when creating ECS services and tasks, improving retry behavior after transient failures.
• [dagster-dbt] DbtCloudComponent now supports an option to include a polling sensor for monitoring dbt Cloud job runs.

Bugfixes

• Fixed snap_to_yaml incorrectly removing empty dicts that represent valid config values for Permissive, Map, and Noneable config types.
• Fixed an issue where resources defined in parent components could incorrectly trigger DagsterInvalidDefinitionError during nested component post_processing resolution. (Thanks, @vidiyala99!)
• [dagster-aws] Fixed PipesECSClient incorrectly treating a task that failed to start as a successful execution.
• [dagster-fivetran] Fixed an incorrect connector URL format generated by the Fivetran integration.
• [ui] Fixed a performance issue where canceling a backfill before any runs executed would unnecessarily fetch the entire asset materialization history.
• [dg] Fixed dg plus deploy not correctly pulling in environment variables when refreshing definitions state for state-backed components.
• [ui] Fixed ANSI color rendering in run logs.
• [ui] Fixed a visual rendering issue with the graph depth input field.

Documentation

• Added a migration guide for moving from sensors to declarative automation.
• Added documentation for the ECS executor.

Dagster Plus

• Alert policies can now be muted for a configurable duration. Muted policies do not send notifications and show when they will resume.
• [dagster-dbt] Added dg plus integrations dbt download-manifest command to download dbt manifests from Dagster Plus for local development.

AI has made contributing to open source easier but reviewing contributions is still hard. At Dagster, we're improving the contributor experience with smarter review tooling, clearer guidelines, and a focus on contributions that are easier to evaluate, merge, and maintain.

New

• Removed the setuptools<82 pin from the dagster package.
• Added support for the partitions attribute in asset selection syntax to filter assets by partition definition type (e.g., partitions:"static").
• [dagster-spark] Added Spark Declarative Pipeline support in feature preview.
• [ui] Added a date range picker to the backfill modal for date-formatted partitions.
• [ui] The "Report evaluation" action for asset checks has been moved to a consolidated dropdown menu.

Bugfixes

• [dagster-dbt] Fixed issue that could cause DbtProjectComponents with custom translation methods to not work with utilities like build_schedule_from_dbt_selection.
• [ui] Fixed asset catalog sidebar not clearing key prefix filters when navigating back to root.
• [ui] Fixed deployment switcher button text color in Light Mode.
• [ui] Fixed duplicate "Catalog" breadcrumb on asset detail pages.
• [ui] Fixed lineage graph edges disappearing when zoomed out by increasing the maximum visible edge limit.

Documentation

• Added cross-account service discovery documentation for ECS agents, including the service_discovery_role_arn configuration parameter.
• Added Snowflake EL ingestion pattern documentation.

New

• Execution context classes (OpExecutionContext, AssetExecutionContext, AssetCheckExecutionContext) now expose a multi_partition_key property that returns a MultiPartitionKey when the current run is a multi-partition run.
• Added Braze and Runpod kind tags. (Thanks, @dragos-pop!)
• [dagster-databricks] The Databricks job run URL is now rendered as a clickable link in the Dagster UI.
• [dagster-dbt] Added DbtCloudComponent for loading dbt Cloud projects as Dagster assets using the Components API.
• [dagster-dbt] The dbt_cloud_assets decorator now supports partitioned assets via the partitions_def parameter.
• [dagster-fivetran] Added a polling sensor for Fivetran observability, which detects externally-triggered syncs and emits materialization events.
• [dagster-fivetran] FivetranWorkspace now supports a retry_on_reschedule option to automatically retry syncs rescheduled by Fivetran due to quota limits, as well as resync operations.
• [dagster-fivetran] The Fivetran translator now includes sync schedule and custom report metadata on connector assets.
• [dagster-k8s] The Dagster Helm chart and Dagster+ Kubernetes Agent Helm chart now support k8sApiCaBundlePath to configure a custom CA certificate path for Kubernetes API communication.
• [dagster-k8s] Code location server Kubernetes Services now support a service_spec_config field for arbitrary Kubernetes Service spec overrides (for example, clusterIP: None for headless services).

Bugfixes

• Fixed an issue where time window partitions with exclusions would sometimes create more runs than needed when using single-run backfills.
• Fixed a "Cannot access partition_key for a non-partitioned run" error that could occur with multi-asset definitions that included non-partitioned assets or asset checks alongside partitioned ones.
• [dagster-aws] Fixed s3_pickle_io_manager failing with dynamic outputs when step keys contain bracket characters in the generated S3 object path.
• [dagster-aws] Fixed an issue in PipesEMRServerlessClient where a custom CloudWatch log group name configured in monitoringConfiguration.cloudWatchLoggingConfiguration.logGroupName was ignored, causing log streaming to always use the default /aws/emr-serverless log group. (Thanks, @kchainani-figma!)
• [ui] Fixed the asset recent updates trend visualization when multiple event types share the same run ID.
• [ui] Fixed text wrapping and spacing in the asset event detail view for long partition names and run titles.

Documentation

• Added a troubleshooting guide for Kubernetes agent network connectivity issues, including TCP keepalive configuration recommendations for Helm chart deployments.

DataOps is about building a system that provides visibility into what's happening and control over how it behaves

New

• Added a "Report Execution" dialog to the asset checks detail view, allowing users to manually record check evaluation results.
• Added database pool configuration options (--db-pool-recycle, --db-pool-pre-ping, and others) to dg dev and dagster dev.
• Added dg plus config view command for inspecting the current CLI configuration.
• [ui] Updated the Usage dialog and Run timeline to display "Jobless asset materializations" (previously "Ad hoc materializations") with a tooltip describing what is grouped within this category.
• [ui] Planned run events are now excluded from the event count shown in the run log filter.
• [dagster-azure] Added component support for dagster-azure, including AzureBlobStorageResourceComponent and ADLS2ResourceComponent for declarative YAML configuration of Azure resources.
• [dagster-databricks] DatabricksAssetBundleComponent is now subsettable at the job level, enabling selective execution of individual Databricks tasks.
• [dagster-databricks] DatabricksAssetBundleComponent now uses the Databricks CLI to resolve variable references to task and job names in bundle templates.
• [dagster-databricks] Databricks jobs are now cancelled when the corresponding Dagster run is terminated in DatabricksWorkspaceComponent.
• [dagster-dbt] dagster-dbt now prefers dbt-core for manifest parsing when it is installed.
• [dagster-gcp] Added BigQueryResourceComponent, GCSResourceComponent, GCSFileManagerResourceComponent, and DataprocResourceComponent for declarative YAML configuration of GCP resources.
• [dagster-gcp] BigQueryIOManager now supports a configurable write_mode parameter (truncate, replace, or append).

Bugfixes

• Fixed an issue where auto-run reexecution would attempt to rerun jobs belonging to already-completed or cancelled backfills.
• Fixed an issue where backfill errors that were subsequently retried would remain incorrectly associated with the backfill.
• dg plus pull env now merges pulled secrets into the existing .env file instead of replacing it, preserving any locally-set variables not present in Dagster Plus.
• [ui] Fixed rendering of newlines in markdown blockquotes in the UI.
• [dagster-databricks] Fixed a KeyError for run_page_url in DatabricksWorkspaceComponent.
• [dagster-databricks] Fixed asset mapping in DatabricksAssetBundleComponent and DatabricksWorkspaceComponent to use job and task key combinations, preventing conflicts when task keys are not unique across jobs.

Dagster Plus

• The Dagster+ agent now automatically redeploys local code servers that enter a failure state when the control plane expects them to be loaded.

Standardizing on Databricks is a smart strategic move, but consolidation alone does not create a working operating model across teams, tools, and downstream systems. By pairing Databricks and Unity Catalog with Dagster, enterprises can add the coordination layer needed for dependency visibility, end-to-end lineage, and faster, more confident delivery at scale.

New

• Added event type filtering to the asset Recent Updates timeline, allowing users to toggle between Materializations and Observations when both event types are present. Filter preference is persisted to local storage.
• psycopg2-binary has been removed as a dependency from dagster-postgres. If you were previously relying on this transitive dependency, you may need to explicitly add psycopg2-binary to your project.
• [ui] The launchpad UI no longer prefills optional resource defaults in the config editor. To see and edit the defaults, click "Scaffold all default config".
• [ui] For op jobs in code locations that have reloaded, the launchpad now always shows a message indicating that the config may be stale.
• [ui] Asset health status indicators now distinguish between loading, unknown, and no data states for improved debugging.
• [ui] The asset overview timeline now correctly displays the most recent event without cutting it off.
• [dagster-dbt] DbtProject and DbtProjectComponent now expose prepare_project_cli_args to allow customizing CLI arguments used to generate the manifest.
• [dagster-looker, dagster-omni, dagster-powerbi, dagster-sigma, dagster-tableau] BI integration assets are now automatically enriched with dagster/table_name metadata. Additionally, dagster-tableau and dagster-looker assets now populate dagster/storage_kind based on the upstream connection type.
• [dagster-sling] The sling package import is now deferred to improve import-time performance.

Bugfixes

• Fixed a bug where Dagster incorrectly called __enter__ on nested resource attributes annotated with dagster.ResourceDependency during parent resource setup. (Thanks, @danielgafni!)
• [ui] Fixed text wrapping and layout for long URIs in asset storage metadata section.

Documentation

• Added documentation for custom metrics in Dagster+ Insights.
• Added and enhanced webhook alerting documentation, including a guide for incident.io integration.

Dagster Plus

• Organization-level timezone settings are now supported, allowing organizations to set a default timezone for users who haven't explicitly set a preference.
• [ui] Asset job run data is now included in Job insights.
• [ui] Code locations with loaded but failed agents now display a warning icon with agent status details in a popover.
• [ui] Added run queue time metric visualization to Job insights.
• [ui] Updated Alert policy list style for improved scannability and legibility.

AI coding agents are changing how data engineers work. This Dagster University course shows how to build a production-ready ELT pipeline from prompts while learning practical patterns for reliable AI-assisted development.

Bugfixes

• Fix bug with inclusion of built webapp in dagster-webserver that caused the Dagster UI to fail to load.

New

• Pool names can now be any non-whitespace character, instead of requiring alphanumbeic characters, slashes, and underscores.
• [dagster-aws] The EcsRunLauncher will now retry a task launch when a RunTask API call fails due to a throttling error in an underlying EC2 API call.
• [dagster-tableau] Tableau workbooks and projects can now be filtered using the workbook_selector and project_selector in TableauComponent.

Bugfixes

• [ui] Fixed issue where saved launchpad config was not used for single-partition asset materializations.
• [ui] Fix console error noise during asset lineage navigation.
• [ui] Fixed an issue where the "Start X Automations" and "Stop X Automations" menu options on the Automations tab were sometimes incorrectly disabled.
• [dagster-dbt] The DbtProject constructor now correctly accepts strings for the target_path parameter.