Model Context Protocol · Open Source

Italian statistics,
one prompt away.

An MCP server that connects your AI assistant to ISTAT — the Italian National Institute of Statistics — so you can discover, query and analyse official data with natural language.

Why it exists

ISTAT publishes thousands of datasets through a powerful but complex SDMX API. This server hides the complexity, caches results aggressively and gives Claude exactly the surface it needs to be useful — no more, no less.

Two-layer cache

In-memory and persistent disk cache. Once a dataflow is explored, every follow-up question is instant.

🛡️

Rate limit friendly

Automatic queuing respects ISTAT limits so your conversations never hit an unexpected error.

🌍

Italian or English

All labels and descriptions are available in both languages, matching the way ISTAT publishes them.

🧩

Works with any MCP client

A few lines of config and ISTAT becomes a first-class tool for the AI assistant you already use.

Eight focused tools

Each tool has a single responsibility — combine them to move from curiosity to a ready-to-read data table.

🔎
discover_dataflows

Find ISTAT datasets by keywords, with smart filtering.

🧭
get_constraints

One call returns every dimension with valid values and labels.

🧱
get_structure

Inspect dimensions and codelists for any datastructure.

🏷️
get_codelist_description

Human-readable labels in Italian or English.

💡
get_concepts

Describe an ISTAT concept by its identifier.

📊
get_data

Fetch actual statistical data as a compact TSV table.

🗺️
get_territorial_codes

Resolve REF_AREA codes: Italy, regions, provinces, municipalities.

⚙️
get_cache_diagnostics

Inspect cache state for debugging and observability.

The three-step workflow

A recommended path that turns a vague question into real numbers.

Discover

Ask your AI a question. It calls discover_dataflows and proposes the most relevant datasets.

Inspect

A single get_constraints call returns all dimensions with valid, labelled values.

Fetch

get_data pulls the exact slice as a compact TSV table, ready for analysis.

Get started in two minutes

Python 3.11+ and uv recommended. Full instructions — and a Windows PowerShell helper — live in the repository.

git clone https://github.com/ondata/istat_mcp_server.git
cd istat_mcp_server
uv sync
Full setup guide