Releases · kedro-org/kedro

1.2.0 Major features and improvements

Added @experimental decorator to mark unstable or early-stage public APIs.
Added support for running multiple pipelines in a single Kedro session run via the --pipelines CLI option and pipeline_names argument in KedroSession.run() method.
Updated the spaceflights-pyspark starter to use the new SparkDatasetV2 integration, enabling local, Databricks-native, and remote Spark execution workflows.

Experimental features

Added experimental llm_context_node and LLMContextNode for assembling LLMs, prompts, and tools into a runtime LLMContext within Kedro pipelines.
Added experimental preview_fn argument to Node class to add support for user-injectable node preview functions.
Added new experimental support-agent-langgraph starter, which supports the above experimental features. This starter contains pipelines that leverage LangGraph for agentic workflows and Langfuse or Opik for prompt management and tracing.

Bug fixes and other changes

Set raise_errors=True in find_pipelines() calls in the project template's pipeline_registry.py to ensure pipeline discovery errors are raised during project runs.
Fixed packaged runs logging the current working directory name; they now log the installed package name (or project path) instead.

Documentation changes

Added beginner-friendly notes on uvx installation.
Updated Databricks deployment docs to cover Spark Connect and Unity Catalog – first workflows, and local-to-remote development.

Community contributions

Many thanks to the following Kedroids for contributing PRs to this release:

Mohmn

1.1.1 Bug fixes and other changes

Fixed project version mismatch error. The error is now only raised when the major version of the project and Kedro package differ, allowing minor and patch version differences without unnecessary failures.

1.1.0 Major features and improvements

Added the ignore_hidden parameter to the OmegaConfigLoader.
Dropped support for Python 3.9 (EOL Oct 2025). Minimum supported version is now 3.10.

Bug fixes and other changes

Bumped click dependency to support versions 8.2.0 and above.
Fixed typos in docs and docstrings.
Fixed duplicate execution of the deindex-old-docs.js script on Kedro documentation pages, preventing double injection of noindex meta tags after the MkDocs migration.

Documentation changes

Added a note on programmatically creating lambdas when lazily saving a PartionedDataset.
Added Dagster as a supported deployment platform and the kedro-dagster plugin to the list of community plugins.
Added new data-validation documentation covering how to use Pandera and Great Expectations with Kedro.

Community contributions

Many thanks to the following Kedroids for contributing PRs to this release:

0.19.15 Bug fixes and other changes

Improved namespace validation efficiency to prevent significant slowdowns when creating large pipelines
Made deprecation warning appear only when the deprecated --namespace flag is used

1.0.0 Major features and improvements

Data Catalog

The previously experimental KedroDataCatalog has been renamed to DataCatalog and is now the default catalog implementation.
It retains the dict-like interface, supports lazy dataset initialisation, and delivers improved performance.
While this change is seamless for users following standard Kedro workflows, it introduces a richer API for programmatic use:
- New pipeline-aware commands, available via both the CLI and interactive environments.
- Simplified handling of dataset factories.
- Centralised pattern resolution via the CatalogConfigResolver property.
- Ability to serialise the catalog to configuration and reconstruct it from it.

Namespaces

Added support for running multiple namespaces within a single session with --namespaces CLI option and namespaces argument in KedroSession.run() method.
Improved namespace validation efficiency to prevent significant slowdowns when creating large pipelines.
Added stricter validation to dataset names in the Node class, ensuring . characters are reserved to be used as part of a namespace.
Added a prefix_datasets_with_namespace argument to the Pipeline class which allows users to turn on or off the prefixing of the namespace to the node inputs, outputs, and parameters.
Changed pipeline filtering for namespace to return exact namespace matches instead of partial matches.

Other features and improvements

Changed the default node name to be formed of the function name used in the node suffixed by a secure hash (SHA-256) based on the function, inputs, and outputs, ensuring uniqueness and improved readability.
Added an option to select which multiprocessing start method is going to be used on ParallelRunner via the KEDRO_MP_CONTEXT environment variable.
Added --only-missing-outputs CLI flag to kedro run. This flag skips nodes when all their persistent outputs exist.
Updated kedro registry describe to return the node name property instead of creating its own name for the node.
Removed pre-commit-hooks dependency for new project creation.

Breaking changes to the API

CLI

kedro catalog create command has been removed.
kedro catalog list, kedro catalog rank, and kedro catalog resolve commands have been replaced with kedro catalog describe-datasets, kedro catalog list-patterns and kedro catalog resolve-patterns commands, respectively.
The kedro run option --namespace has been removed and replaced with --namespaces.
The kedro micropkg CLI command has been removed as part of the micro-packaging feature deprecation.

API

Private methods _is_project and _find_kedro_project are changed to is_kedro_project and find_kedro_project.
Renamed instances of extra_params and _extra_params to runtime_params.
Removed the modular_pipeline module and moved functionality to the pipeline module instead.
Renamed ModularPipelineError to PipelineError.
Pipeline.grouped_nodes_by_namespace() was replaced with group_nodes_by(group_by), which supports multiple strategies and returns a list of GroupedNodes, improving type safety and consistency for deployment plugin integrations.
Renamed session_id parameter to run_id in all runner methods and hooks to improve API clarity and prepare for future multi-run session support.
Removed the following DataCatalog methods: _get_dataset(), add_all(), add_feed_dict(), list(), and shallow_copy().
Changed the output of runner.run() and session.run() — it now always returns all pipeline outputs, regardless of catalog configuration.
Removed the AbstractRunner.run_only_missing() method, an older and underused API for partial runs. Please use --only-missing-outputs CLI instead.

Documentation changes

Revamped the look and feel of the Kedro documentation, including a new theme and improved navigation with mkdocs as the documentation engine.
Updated the DataCatalog documentation with improved structure and detailed description of new features. Read the DataCatalog documentation here.

Community contributions

Many thanks to the following Kedroids for contributing PRs to this release:

Migration guide from Kedro 0.19.* to 1.*

See the migration guide for 1.0.0 in the Kedro documentation.

1.0.0rc3

Major features and improvements

Changed DataCatalog.__getitem__ to raise DatasetNotFoundError for missing datasets, aligning with expected dictionary behavior.

Bug fixes and other changes

Breaking changes to the API

Upcoming deprecations for Kedro 1.0.0

Documentation changes

Community contributions

1.0.0rc2

Major features and improvements

Added --only-missing-outputs CLI flag to kedro run. This flag skips nodes when all their persistent outputs exist.
Removed the AbstractRunner.run_only_missing() method, an older and underused API for partial runs. Please use --only-missing-outputs CLI instead.

Bug fixes and other changes

Improved namespace validation efficiency to prevent significant slowdowns when creating large pipelines

Breaking changes to the API

Upcoming deprecations for Kedro 1.0.0

Documentation changes

Community contributions

1.0.0rc1

Major features and improvements

Added stricter validation to dataset names in the Node class, ensuring . characters are reserved to be used as part of a namespace.
Added a prefix_datasets_with_namespace argument to the Pipeline class which allows users to turn on or off the prefixing of the namespace to the node inputs, outputs, and parameters.
Changed the default node name to be formed of the function name used in the node suffixed by a secure hash (SHA-256) based on the function, inputs, and outputs, ensuring uniqueness and improved readability.
Added an option to select which multiprocessing start method is going to be used on ParallelRunner via the KEDRO_MP_CONTEXT environment variable.

Bug fixes and other changes

Changed pipeline filtering for namespace to return exact namespace matches instead of partial matches.
Added support for running multiple namespaces within a single session.
Updated kedro registry describe to return the node name property instead of creating its own name for the node.

Documentation changes

Updated the DataCatalog documentation with improved structure and detailed description of new features.

Community contributions

Breaking changes to the API

Private methods _is_project and _find_kedro_project are changed to is_kedro_project and find_kedro_project.
Renamed instances of extra_params and _extra_params to runtime_params.
Removed the modular_pipeline module and moved functionality to the pipeline module instead.
Renamed ModularPipelineError to PipelineError.
Pipeline.grouped_nodes_by_namespace() was replaced with group_nodes_by(group_by), which supports multiple strategies and returns a list of GroupedNodes, improving type safety and consistency for deployment plugin integrations.
The micro-packaging feature and the corresponding micropkg CLI command have been removed.
Renamed session_id parameter to run_id in all runner methods and hooks to improve API clarity and prepare for future multi-run session support.
Removed the following DataCatalog methods: _get_dataset(), add_all(), add_feed_dict(), list(), and shallow_copy().
Removed the CLI command kedro catalog create.
Changed the output of runner.run() — it now always returns all pipeline outputs, regardless of catalog configuration.

Migration guide from Kedro 0.19.* to 1.*

See the migration guide for 1.0.0 in the Kedro documentation.

0.19.14 Major features and improvements

Added execution time to pipeline completion log.

Bug fixes and other changes

Fixed a recursion error in custom datasets when _describe() accessed self.__dict__.

Community contributions

Many thanks to the following Kedroids for contributing PRs to this release:

Yury Fedotov

0.19.13 Major features and improvements

Unified pipeline() and Pipeline into a single module (kedro.pipeline), aligning with the node()/Node design pattern and improving namespace handling.

Bug fixes and other changes

Fixed bug where project creation workflow would use the main branch version of kedro-starters instead of the respective release version.
Fixed namespacing for confirms during pipeline creation to support IncrementalDataset.
Fixed bug where OmegaConfcause an error during config resolution with runtime parameters.
Cached inputs in Node when created from dictionary for better performance.
Enabled pluggy tracing only when logging level is set to DEBUG to speed up the execution of project runs.

Upcoming deprecations for Kedro 1.0.0

Added a deprecation warning for catalog CLI commands. The following commands will be replaced with their alternatives - kedro catalog rank, kedro catalog list, kedro catalog resolve and the kedro catalog create command will be removed.
Added a deprecation warning for KedroDataCatalog that will replace DataCatalog while adopting the original DataCatalog name.
Add deprecation warning for --namespace option for kedro run. It will be replaced with --namespaces option which will allow for running multiple namespaces together.
The modular_pipeline module is deprecated and will be removed in Kedro 1.0.0. Use the pipeline module instead.

Note: On March 20th, a security vulnerability, CVE-2024-12215, was identified in Kedro. This issue stems from the deprecated micropackaging functionality, which is scheduled for removal in the upcoming Kedro 1.0 release. While we agree with the CVE assigned, this vulnerability only poses a risk if you pull a malicious micropackage from an untrusted source. If you're concerned, we recommend avoiding the micropackaging feature for now and upgrading to Kedro 1.0 once it's released.

Documentation changes

Updated Dask deployment docs.
Added non-jupyter environment integration page (for example Marimo) with dynamic Kedro session loading.

Community contributions

Many thanks to the following Kedroids for contributing PRs to this release: