Sweden's data portal

Finding out whether public data on a subject exists at all, without knowing which agency would have published it.

What it is

dataportal.se is Sweden’s national catalogue of open data — the index of what public bodies have published, not the data itself. This integration searches it two ways: against a locally maintained index for semantic search, and live against the portal’s own API for the most current answer.

Use it for

• “Which open datasets exist on cycling infrastructure?”
• “Has any municipality published data on this, and under what licence?”
• “Who publishes the most datasets in this theme?”
• “Find datasets about air quality, and tell me which are actually machine-readable.”
• “What is in this dataset, and where is the file?”

Tools

Function	Answers
Usage guide	When to search the index and when to go live — the assistant reads this first
Dataset search	Datasets matching a question, searched semantically over the index
Live search	The same question against the portal’s current API
Get dataset	What a single dataset holds, its publisher, licence and distributions
Organisations	Which bodies publish to the portal
Themes	The catalogue’s own subject classification
Statistics	How much the catalogue holds, and in what shape

Data and permissions

• Open data, no key. Nothing about your organisation passes through.
• Read-only.
• A catalogue, not a warehouse. The integration tells you a dataset exists and where it is; it does not download and analyse the data behind it.

Getting started

Add the MCP server to an assistant and ask.

Limitations

• It indexes descriptions, not contents. A dataset whose description is vague is hard to find no matter how good the data is, and the portal is only as good as what publishers wrote.
• The local index has a lag. For “is this published yet”, the live search is the one to use, and the usage guide says so.
• Sweden only.
• Cataloguing is not availability. A listed distribution can still be a dead link on the publisher’s side.