Magda is a federated, open-source data catalogue for both big and small data. It gives a single view of data regardless of where it is stored or where it came from, by combining metadata from sources such as CKAN, data.json, Excel or CSV inventories and REST APIs into one search index. Magda offers hybrid search (vector and BM25), an extensible "minion" framework that checks links, normalises formats and calculates quality, previews of tables and maps, and authorisation built on Open Policy Agent. The software runs on Kubernetes and is installed with Helm, in the cloud, on premises or locally.
The project was started by CSIRO's Data61 and the Australian Department of the Prime Minister and Cabinet, and was built for data.gov.au, which it served from 2019 to June 2025. Development continues in the open on GitHub under the Apache 2.0 licence. Because you host Magda yourself, your organisation decides where the catalogue and its metadata reside. The project originates in Australia. No vendor-hosted cloud is required.