odd-great-expectations

Great Expectations checkpoint ValidationAction that pushes expectation results to the ODD Platform.

Status: Stable, V3 API only. No Great Expectations Cloud Solution support; no V2 API support.

odd-great-expectations is a push adapter for Great Expectations. It plugs in as a ValidationAction on a GE checkpoint — every time the checkpoint runs, the action serialises the validation results into the ODD specification and pushes them to the ODD Platform. The result: GE expectation outcomes show up alongside other data-quality signals in the catalog.

For the broader pull-vs-push picture, start at the Integrations hub.

Requirements

• Great Expectations V3 API. The action targets the V3 checkpoint mechanism (great_expectations.checkpoint.Checkpoint).

• One of the supported execution engines:

  • SQLAlchemy engine.

  • Pandas engine.

The V2 API and the GE Cloud Solution are not supported.

Capability	Supported
GE V3 API	✓
SqlAlchemyEngine	✓
PandasEngine	✓
GE V2 API	✗
GE Cloud Solution	✗

Source: odd-great-expectations README → Supporting.

Installation

pip install odd-great-expectations

The package exposes odd_great_expectations.action.ODDAction, a class derived from GE's ValidationAction.

Configuration

Add ODDAction to your checkpoint's action_list:

name: <CHECKPOINT_NAME>
config_version: 1.0
template_name:
module_name: great_expectations.checkpoint
class_name: Checkpoint
run_name_template: '%Y%m%d-%H%M%S-my-run-name-template'
expectation_suite_name:
batch_request: {}
action_list:
  # ...other actions
  - name: store_metadata_to_odd
    action:
      module_name: odd_great_expectations.action
      class_name: ODDAction
      platform_host: http://odd-platform.internal:8080
      platform_token: <COLLECTOR_TOKEN>
      data_source_name: my_local_qa_test
evaluation_parameters: {}

Action field	Value
module_name	odd_great_expectations.action (literal).
class_name	ODDAction (literal).
platform_host	ODD Platform URL.
platform_token	Collector token issued by the platform under Management → Collectors.
data_source_name	Unique name for the GE data source as it appears in the platform — e.g. local_qa_test.

Run the checkpoint as you normally would:

great_expectations checkpoint run <CHECKPOINT_NAME>

The ODDAction runs on every checkpoint execution after the validation completes; success / failure / partial-success outcomes are all pushed.

What gets sent

• Quality test results for every expectation in the checkpoint.

• The dataset(s) the expectations apply to (if not already in the catalog, they are registered under the supplied data_source_name).

• The expectation outcome status (pass / fail) and any GE result payload that maps cleanly onto the ODD quality-test schema.

The platform surfaces these on the dataset's detail page and aggregates them into the Quality Dashboard.

Known limitations

• V3 API only. GE V2 API checkpoints are not compatible.

• No GE Cloud Solution support. The push happens from the Python process running the checkpoint; the action does not integrate with GE Cloud's hosted execution.

• Two engines. Outside SqlAlchemyEngine and PandasEngine (e.g. SparkDFExecutionEngine), the action is not exercised — file an issue if you need other engines.

• platform_token is plain text in the checkpoint YAML by default. Source it from a secret store / environment via your YAML loader if your deployment requires non-plaintext config.

Where to next

• Quality Dashboard — where GE results aggregate alongside other quality signals.

• odd-dbt — paired adapter when GE expectations and dbt tests cover overlapping datasets.

• Repo — sources, releases, and issues.