IntervaOpen the desk

API and MCP

The catalog is open to programs as well as people. Reading datasets, series and search needs no key. Briefs, comparisons, development finance and Ask are part of a plan.

MCP server

Give Claude, ChatGPT or any MCP client the catalog as tools: find_data, country_overview, get_series, compare_sources and cite_dataset. Streamable HTTP, no key.

https://interva-api.nq5kq4f69b5zj.eu-west-3.cs.amazonlightsail.com/mcp

In a client that reads a JSON config:

{
  "mcpServers": {
    "interva": {
      "type": "http",
      "url": "https://interva-api.nq5kq4f69b5zj.eu-west-3.cs.amazonlightsail.com/mcp"
    }
  }
}

In Claude Code:

claude mcp add --transport http interva https://interva-api.nq5kq4f69b5zj.eu-west-3.cs.amazonlightsail.com/mcp

REST API

JSON over HTTPS from https://interva-api.nq5kq4f69b5zj.eu-west-3.cs.amazonlightsail.com. The full reference is at /docs.

  • GET /discover?q=Kenya GDP 2015-2024

    Search that reads a question: datasets for the subject, and the named country’s figures.

  • GET /datasets?limit=50&offset=0

    Every dataset, a page at a time.

  • GET /datasets/{id}

    One dataset: publisher, licence, provenance, citation and its series by country.

  • GET /datasets/{id}/download

    The whole dataset as CSV: geography, country, period, value, unit, flags.

  • GET /series/{id}

    One country’s series, every observation.

  • GET /countries/{code}

    A country’s headline figures, topics and sources.

  • GET /geographies

    Every country and aggregate, with region and income group.

  • GET /sources

    The publishers, and what each contributes.

curl "https://interva-api.nq5kq4f69b5zj.eu-west-3.cs.amazonlightsail.com/discover?q=Kenya%20GDP%202015-2024"
curl -o gdp.csv "https://interva-api.nq5kq4f69b5zj.eu-west-3.cs.amazonlightsail.com/datasets/worldbank-gdp-current-usd/download"

Citing and licences

Every figure keeps the licence its publisher gave it, most often CC BY 4.0; each dataset states its own. Cite the original publisher first and Interva as where you retrieved it. Each dataset carries a ready-made citation, and the cite_dataset tool returns it. A catalog-wide description is at /catalog.jsonld (DCAT), and a summary for language models at /llms.txt.