# publicdata.au [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/National-Digital/publicdata.au  
**GitHub Stars:** 4  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/publicdata-au

## Description
Search, filter, count and sum Australian government open data, with every version kept.

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What publicdata.au MCP server does

The publicdata.au MCP server provides an agent-facing way to search and analyze Australian government open data. Its supported operations include finding datasets, applying filters, counting matching records, and calculating sums. The underlying publicdata.au service republishes datasets from government sources and keeps each publisher release as a separate dated version.

Datasets are available at URLs intended to remain stable. Depending on the source and publication, the service can provide CSV, Excel, JSON, Parquet, SQLite, DuckDB, GeoJSON, and GeoPackage files. This makes the catalogue useful for both conversational investigation and workflows that need to hand a file to another data tool.

The service is independently operated by National Digital. It is not run, funded, or endorsed by an Australian government agency. Dataset licensing remains with the original publisher, and use may require the attribution specified by that licence.

## How it works

The publicdata.au MCP server sits over publicdata.au’s published catalogue and query functionality. An agent can use dataset-level operations to locate relevant government data, narrow results with filters, and obtain counts or sums without first downloading and processing every file locally.

Every release made by a publisher is preserved as a version. Alongside the retained content, publicdata.au records the schema, provenance, the publisher’s original file, and a comparison with the preceding release. This version history helps an agent distinguish current data from an earlier publication and investigate changes over time.

The project’s site includes a query API, while its build pipeline fetches and prepares source data for publication. The repository also contains Python and R client packages that accept dataset slugs, although the provided material does not specify whether those clients are required for MCP use.

## Tools and capabilities

The documented data operations support these tasks:

- Search the Australian government open-data catalogue.
- Filter records or dataset results according to a query.
- Count matching records.
- Sum values from matching data.
- Work with retained historical releases.
- Retrieve data in common tabular, database, and geospatial formats.

These capabilities suit questions such as identifying relevant datasets, checking how many records meet a condition, or totaling a numeric field after filtering. The material does not define the exact MCP tool names, argument schemas, pagination behavior, or supported aggregation rules.

## Setup and configuration

The repository states that a working copy can be set up and used without credentials. Its contribution workflow builds fixtures, fetches a real dataset, and demonstrates routine changes. The provided documentation does not include an MCP-specific installation command, transport configuration, endpoint URL, or required environment variables, so those details should be confirmed before deployment.

The project’s automated deployment uses Cloudflare Pages and R2, but those infrastructure credentials belong to the repository’s deployment workflows. They are not documented as requirements for an agent connecting to the public data service.

## Limitations and notes

Coverage depends on the datasets that publicdata.au has registered and processed; the material does not promise that every Australian government dataset is available. Source data remains subject to each publisher’s licence, and publicdata.au’s register text is released under CC BY 4.0.

The repository code is licensed under AGPL-3.0-or-later. The Python and R clients use MIT licences stated in their individual packages. No specific AI client compatibility is named in the provided material, and no claim can be made here about support for a particular desktop or editor client.

_Full upstream README: https://allmcps.com/mcp/publicdata-au/readme_