Apiverket MCP

What vinvuk/apiverket-mcp MCP server does

The vinvuk/apiverket-mcp MCP server connects MCP-compatible AI clients to Swedish public data exposed by Apiverket. It provides endpoint discovery before data retrieval, helping an agent identify supported paths, required parameters, and the relevant data category instead of guessing API routes.

The catalog contains 139 endpoints in 16 categories. Supported areas include weather, transport, economy, health, environment, parliament, police, education, municipalities, jobs, culture, tourism, geography, and company data. Returned API results use JSON with a metadata envelope and a data object.

Three MCP tools are documented:

• govdata_discover searches endpoint names, descriptions, paths, and categories. With no arguments, it lists available categories and endpoint counts.
• govdata_query calls a discovered Apiverket path with optional path and query parameters.
• govdata_account shows sanitized key information, account tier, daily usage, company-search and company-lookup quotas, reset timing, and relevant limits.

How it works

The vinvuk/apiverket-mcp MCP server runs locally over stdio. Discovery uses the built-in endpoint catalog without making an API request. Query operations send the selected path and parameters to Apiverket using the configured API key. Structured failures are returned as recovery guidance, including cases where the API is unreachable, a path is unsupported, or parameters are invalid.

A typical workflow is:

1. Search with govdata_discover using a topic such as weather, train departures, pension, or company data.
2. Read the returned endpoint path, parameters, and description.
3. Call govdata_query with the required path and query values.
4. Use pagination guidance if the response is truncated.
5. Check govdata_account after rate limits or when quota and tier details matter.

Company workflows distinguish searching by name from looking up a known organisation number. Search results can provide an org_number for later lookup, while the documented company API does not expose board, officer, owner, or UBO subresources.

Setup and configuration

Use Node.js 18 or newer. The documented MCP client configuration starts the package with npx -y apiverket-mcp-server. Configure these variables:

• GOVDATA_API_URL: Apiverket base URL; the documented default is https://apiverket.se.
• GOVDATA_API_KEY: API credential; sk_test_demo is available for sandbox setup, while live keys are used for production data.

The repository can also be built from source with npm install followed by npm run build after cloning it. The project is licensed under MIT.

Tools and capabilities

The vinvuk/apiverket-mcp MCP server supports endpoint search by keyword or category, direct calls with path parameters such as city or station, and query parameters such as search terms, coordinates, limits, and offsets. Example data tasks include retrieving Stockholm weather, train departures, exchange rates, municipal population, police events, electric-vehicle chargers, job listings, and company records, provided the corresponding endpoint is discovered first.

Limitations and notes

An Apiverket API key is required, although the documented sandbox key can be used to verify client setup. Daily limits and company-search or company-lookup quotas vary by key tier. Agents should not infer undocumented paths; discovery is the supported way to select endpoints and understand required parameters. Large responses may be truncated, with pagination instructions returned to the client.