TrustyData MCP Server

Connect your LLM (Claude, ChatGPT, IDE assistants…) to TrustyData — French address data quality, geocoding, routing, company lookup (SIRENE) and catchment-area statistics (INSEE), built on official open data sources (BAN, INSEE, SIRENE, IGN, OpenStreetMap).

▶️ Try it live — public demo agent, no account needed Ask a real agent to verify an address, list communes, look up a company or compute a route: it calls this MCP server in front of you.

This is a hosted, remote MCP server — there is nothing to install or run. Point your MCP client at the endpoint below and sign in with your TrustyData account.

This repository hosts the public listing metadata for the TrustyData MCP server (server.json, documentation, examples). The server itself is a hosted service — the source of the underlying data API is not part of this repo.

What it does

TrustyData turns messy French addresses into clean, authoritative data and adds geographic context, directly inside your LLM conversation:

• Verify & normalize a French postal address against the official BAN reference (the result is authoritative — an empty result means no match).
• Search addresses and localities by name, postal code or INSEE code, with optional filters (department, region, population).
• Proximity search — find addresses or points near a location.
• Routing — compute a road route or a travel-time/distance matrix in France (OpenStreetMap), by car, on foot or by bike.
• Company lookup — search French companies and establishments in the official SIRENE registry and get full records (executives, finances, collective agreements, depending on your plan).
• Catchment areas — population, households, income, age and socio-professional profile, spending potential of a drive-time or radius zone (INSEE Filosofi & census); compare up to 10 zones and measure what each one covers exclusively.
• Activity mix — how many SIRENE establishments a catchment area holds, by family and sub-family of activity, with a density per 10,000 inhabitants and an index against a reference territory: France by default, or the zone's dominant département when you ask for it. A floor, never a total: only active, geolocated establishments are counted, and the response says which reference was used and what share of it that count represents.

Richer fields (e.g. INSEE Filosofi statistical grid, Lambert 93 coordinates) are returned depending on your plan.

Tools

All tools are advertised to every client. If your plan doesn't cover a tool, it returns an actionable upgrade message instead of failing silently.

Prompts

Beyond tools, the server ships five prompts — guided starters that fill in the right parameters for you. In Claude Code they show up as /mcp__trustydata__<name>; other MCP clients surface them their own way.

Prompt names and wording are French, like the data they describe. A prompt returns a message, not a result: it calls no tool by itself and consumes no quota — the tool it points at applies its own plan requirement when the assistant calls it.

Connect

Claude (claude.ai / Claude Desktop)

Add a custom connector pointing to:

https://mcp.trustydata.app/mcp

You'll be prompted to sign in via OAuth with your TrustyData account the first time a tool is used.

Generic MCP client (Streamable HTTP)

{
  "mcpServers": {
    "trustydata": {
      "url": "https://mcp.trustydata.app/mcp"
    }
  }
}

Don't have an account yet? Start free on trustydata.fr — the Discovery plan is free (5,000 requests/month, no card required).

Example prompts (free-form)

No prompt needed — just ask:

Vérifie et normalise cette adresse : "1 rue de Rivol 75001 Pari"

Quelle est la population de la commune de Bourg-en-Bresse ?

Trouve les adresses proches du 2 avenue de la Gare à Annecy.

Calcule l'itinéraire routier entre Lyon Part-Dieu et l'aéroport Saint-Exupéry.

Trouve les entreprises de plomberie à Annecy et donne-moi la fiche du premier établissement.

Combien de personnes habitent à 15 minutes en voiture du 1 rue Scribe, 75009 Paris, et quel est leur niveau de vie ?

Compare 10 minutes en voiture autour de la Part-Dieu et autour de Bellecour : quelle zone capte le plus d'habitants qu'elle est seule à couvrir ?

Combien d'établissements de restauration y a-t-il à 10 minutes à pied de la place Bellecour, et est-ce beaucoup pour le nombre d'habitants ?

Data sources & attribution

TrustyData relies on official, regularly-updated open data. Please keep the attributions returned by the tools:

• Addresses & communes — Base Adresse Nationale (BAN) & INSEE
• Companies — SIRENE (INSEE), Registre National des Entreprises (INPI)
• Statistical context & catchment areas — INSEE Filosofi and census, IGN Contours IRIS
• Real-estate sale prices — DGFiP, Demandes de valeurs foncières (Etalab, open licence)
• Establishments by activity — SIRENE (INSEE), active geolocated establishments only
• Routing — OpenStreetMap contributors (ODbL)

Documentation

Full API guides, endpoint reference and examples: https://trustydata.fr/docs/

License

The contents of this listing repository (documentation, server.json, examples) are released under the MIT License. The hosted service and the underlying data API are operated by TrustyData and governed by its terms.