Connect data to Great Expectations with AI
Set up a readable batch from a file collection, database table or dataframe, with a small live read to confirm the selection.
You can connect data to Great Expectations with AI when you know where a feed lives but need a repeatable way to select it. A set of delivery files, for instance, might arrive in one folder with the reporting month embedded in each filename. The gx-configure-data-source skill sets up the connection, identifies the relevant files and defines how one run picks its batch.
The final check opens the data itself. That matters when a directory exists but the naming pattern picks nothing. The agent reports which objects it created or reused and whether the configuration will survive the session.
What it does
- Read a single monthly partition before writing rules against it.
- Target a database table through an authorised connection.
- Use a dataframe that already exists in the current Python session.
- Keep credentials in environment variables while configuring a connection.
- Distinguish an inaccessible source from an empty or incorrectly selected batch.
Install and use
- Prepare the current GX Core package in Python and install the skill in a supported coding agent. Remote sources also need their usual driver and credentials.
- Describe the feed, the batch you want and which sample rows the agent may view.
- Review the read result and choose whether an in-memory setup should be saved to a project directory.
FAQ
Source
Made by Great Expectations. Original skill: gx-configure-data-source, published on GitHub under the Apache-2.0 licence.
View on GitHub