StatFin MCP: Finland's 3,000+ Official Tables, No API Key Needed

Finland's Official Stats Engine, Now in Your AI Workflow

You want authoritative population data for Finland. Or housing prices. Or labor market trends broken down by region and year. The old way: navigate Statistics Finland's website, find the right table, figure out the query builder, export a CSV, import it somewhere useful.

The new way: ask your AI assistant — and let io.github.3121n/statfin-mcp do the heavy lifting.

This MCP server plugs directly into the StatFin database — Finland's official national statistics repository — through the PxWeb API. No API key. No bureaucratic signup. Just 3,000+ tables of government-grade data, instantly accessible.

The server exposes four tightly scoped tools that mirror exactly how you'd want to explore a statistical database: discover, inspect, filter, fetch.

That's it. No bloat. No abstraction layers doing weird things behind your back. Each tool has a clear job.

All tools support three languages out of the box — English, Finnish (fi), and Swedish (sv) — which is appropriate given Finland's bilingual official status. A small detail, but it signals that the author thought carefully about the actual users of this data.

1. sok_tabell — Search the Catalog

Start here. Feed it keywords, get back the most relevant tables with their tabell_id. The smart part: it matches against a bundled local index rather than firing a live API call for every search. That means fast, offline-capable discovery.

Query "unemployment rate" or "birth rate by municipality" — it finds the right table IDs so you can proceed.

2. list_emner — Browse Subject Areas

When your keyword search misses — because you don't know the right terminology — this is your fallback. It lists approximately 140 subject areas (topics like vaerak for Population Structure) that organize the entire StatFin catalog.

Think of it as the table of contents for Finland's statistical universe.

3. hent_tabell — Read the Metadata

Once you have a tabell_id, this tool fetches the table's full metadata: title, dimensions, valid value codes, and labels. This is the step most people skip and then wonder why their data query fails.

You need to know what filters are valid before you query. This tool tells you.

4. hent_data — Pull the Numbers

The payoff. Submit a tabell_id plus a filter map, get back JSON-stat2 — a structured statistical data format that's well-suited for programmatic use. There's also a top filter shortcut for grabbing the latest time period without knowing the exact year code.

Total: 92/100. That's a strong score — and notably, it's built on genuine technical decisions, not marketing.

The target audience is narrower than a general-purpose data server — but that's a feature, not a bug.

Researchers and analysts working on Finnish demographics, economics, or social policy now have a direct AI-native interface into official statistics. No more manual table hunting.

Journalists and data reporters covering Finland can query historical trends conversationally — "show me Helsinki housing prices for the last five years" — and get structured data back without an intermediate export step.

Developers building applications on top of Finnish government data get a clean MCP abstraction over the PxWeb API, with proper metadata inspection built in. The JSON-stat2 output is already formatted for programmatic consumption.

And for anyone curious about what MCP servers should look like: a focused domain, clean tool separation, sensible output formats, and zero unnecessary dependencies. This is a good example to study.

The zero-star GitHub count is the only real yellow flag — adoption is early, and community support is unproven. But the server itself is technically solid and solves a real problem elegantly.

Finland's official statistics belong in your AI workflow. This server makes that happen.