Datacube Sk

What Datacube Sk does

Datacube Sk exposes Slovak statistical cubes from DATAcube, the Statistical Office of the Slovak Republic's data service. Its three pack-specific tools support the full path from finding a cube to retrieving observations: search the collection, inspect the cube's dimensions, and submit a positional data query.

The integration is unofficial and is not affiliated with, endorsed by, or published by the upstream statistical office. Results are returned in machine-readable formats suited to agent workflows, including JSON-stat for data responses.

How it works

A typical query begins with list_datasets. Supply an optional case-insensitive substring from an English dataset label to narrow the collection. Each match includes a cube code, label, last update date, and ordered dimension codes.

Pass the selected cube code to dataset_dimensions. This returns the dimensions in path order, along with the valid value codes and human-readable labels for each one. That order matters: get_data expects one selection for every dimension, using exactly the sequence returned by this tool.

For each dimension, get_data accepts one value code, several comma-separated codes, or all. The response uses JSON-stat: value contains the flattened data array, while dimension, size, and id describe how the values are arranged.

Setup and configuration

The hosted MCP endpoint is:

https://gateway.pipeworx.io/datacube-sk/mcp

An MCP client can register that URL under a datacube-sk server entry. The endpoint also exposes shared Pipeworx meta-tools in addition to the pack-specific tools, so its tools/list result is larger than the three tools described here. The gateway's ask_pipeworx tool can route natural-language requests across the broader catalog, including this data source.

For a connection containing only the Datacube Sk tools, run the npm package locally over standard input and output:

{
  "mcpServers": {
    "datacube-sk": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-datacube-sk"]
    }
  }
}

The package can also be started directly with npx -y @pipeworx/mcp-datacube-sk. No Pipeworx account is required for the initial hosted HTTP calls, and the standalone mode does not use a gateway account or gateway round-trip.

Tools and capabilities

• list_datasets searches DATAcube by a case-insensitive substring of the English label and returns identifiers, labels, update dates, and dimension-code order.
• dataset_dimensions describes a cube's dimensions and their allowed codes and labels.
• get_data retrieves one cube selection as JSON-stat, supporting individual values, comma-separated value lists, or all values for each dimension.

The hosted service can also be called through the Pipeworx HTTP tools API. For example, dataset search uses a POST request to https://gateway.pipeworx.io/v1/tools/datacube_sk_list_datasets with a JSON query field.

Limitations and notes

get_data requires the number of selections to equal the cube's dimension count. Selections are positional rather than named, so skipping the dimension-discovery step can produce incorrectly formed requests or misaligned results. Cube and value codes should be taken from the discovery tools instead of inferred from labels.

The standalone process exposes only this pack's tools. The hosted pack endpoint adds shared gateway meta-tools, while the full Pipeworx gateway lists tools from all available packs directly. Both hosted URLs use the same gateway and can reach the broader catalog through routing.