MCP + REST API · DVF + DPE + BDNB
DVF·déc. 2025DPE·janv. 2026BDNB·févr. 2026

French Real Estate DVF Data for Your AI Agents

Query 18.1 million French property transactions via REST API or MCP. Connect Claude or ChatGPT in one click, or use curl, Claude Code, Cursor, Cline and your backends.

Use Cases:_
18,1M+ transactions·36 000+ communes·<200 ms
Quick Snippet
Claude Desktop / Cursor / Cline / Windsurflocal
npx -y @normi/mcp-dvf
View access methods

Why Normi DVF

11 years of French real estate data, one prompt away

18.1 million DVF transactions, cleaned, geocoded, and accessible via REST API or MCP. Actual sales — not estimates, not listings.

Plug and play DVF

REST API + STDIO + HTTP streamable for all your tools: Claude Desktop, Cline, Cursor, VSCode, backends and curl scripts.

Advanced market analysis

Comparables, price trends, neighborhood heatmap and address history for actionable insights.

Cleaned data, not raw

~6% bulk sales filtered, ~3% outliers excluded, address-level geocoding. Raw DVF has noise — Normi removes it.

DVF transactions per year (thousands)

201420152016201720182019COVID2020record2021202220232024

Source: DVF, data.gouv.fr, 2014-2024

18,1M+

Geocoded transactions

2014-2025 · 98 départements

36 000+

Communes covered

Metropolitan France

< 200 ms

Median API response

Typical P50 query

In Practice

Your data, at work

Real prompts. Real outputs. Authentic DVF data.

  • Estate agent

    Valuations & comparables

    find_property_comparables

    “Find comparables for a 65m² two-bedroom flat in Lyon 6e and estimate the median price.”

    AddressArea€/m²
    Lyon 6e · 110m67 m²5 045 €
    Lyon 3e · 820m62 m²4 661 €
    Lyon 7e · 950m69 m²4 377 €

    4 853 €/m²

    4 properties found · Lyon · 89ms

    ✓ Ref. Lyon 6e · 63 m² · 315 000 €

  • Investor

    Market analysis

    analyze_price_trends

    “Price per m² in Bordeaux from 2019 to 2024.”

    20192024

    +16 %

    3 500 → 4 050 €/m²

    Bordeaux · flats · 2019–2024

  • Data analyst

    Studies & reports

    compare_locations

    “Compare Vincennes, Montreuil, Pantin and Bagnolet.”

    Vincennes
    6 240 €/m²
    Montreuil
    4 180 €/m²
    Pantin
    3 950 €/m²
    Bagnolet
    3 710 €/m²

    6 240 €/m²

    Vincennes

    East Paris · 2024 · flats

  • Developer

    REST & MCP integration

    search_property_transactions

    commune: "PARIS 15", type_local: "Appartement", surface: 55–75 m²

    DVF response

    {
      "total": 47,
      "median_prix_m2": 10240,
      "_credits": { "used": 5 }
    }

    47

    transactions · 10 240 €/m²

    33 MCP tools · HTTP Streamable · STDIO · REST API

  • EPC data

    ADEME energy ratings × DVF

    analyze_dpe_price_and_thermal_risk

    “What is the price premium for an A-rated home versus a G in Paris?”

    A
    12 400 €/m²
    C
    10 800 €/m²
    E
    9 600 €/m²
    G
    8 500 €/m²

    +46 %

    Green premium, A vs G

    Paris · flats · 2024 · median price/m²

  • BDNB buildings

    32M CSTB buildings × 16.8M DVF transactions

    analyze_building_age_price_impact

    “Do 1970s properties sell for less than Haussmann-era ones in Paris?”

    Before 1914
    11 800 €/m²
    1949–1974
    8 900 €/m²
    1975–1999
    9 200 €/m²
    Since 2000
    10 100 €/m²

    32M

    buildings covered — all of mainland France

    Paris · flats · 2024 · median price/m² by construction period

Scroll sideways to move faster

Quick Installation

3 steps to get started

Free token, JSON config, first DVF query.

Generate Token
  1. 01

    1. Get a token

    Generate a free token (normi_xxx) on the get-token page. Used in NORMI_API_KEY (STDIO) or Authorization Bearer (HTTP).

  2. 02

    2. Configure your MCP client

    STDIO for local clients (Claude Desktop). HTTP streamable for Cline/Cursor/VSCode and your backends (Authorization Bearer).

  3. 03

    3. Test a DVF query

    Ask: Find apartments sold between 2023 and 2024 in Lyon between 200k and 400k.

OAuth
  • One click from the Claude connector directory
  • claude.ai, Claude Desktop, mobile and Cowork
  • Revocable from the dashboard (Connected apps)
// Claude connector directory:
https://claude.ai/directory/normi

// Add → sign in to Normi → Authorize
// No key to copy

// Without the directory: custom connector
https://mcp.normi.fr/mcp

Three guided prompts are included so you can explore the data without reading the docs. See the MCP prompts

DVF & EPC data

Property prices by city

Browse the official DVF database from the DGFiP for 500+ communes — median price per m², EPC green premium, energy-inefficient homes.

500+ communes available · See an example

AI Assistants

Ask your AI assistant about French property prices

Normi is indexed by Claude, ChatGPT, Perplexity and all major AI crawlers. Whether you are researching French real estate or building an AI agent, Normi provides authoritative DVF data your AI can cite.

Claude Desktop / Code

MCP

Install the MCP server once, then ask Claude directly: "What is the median price per m² in Lyon 3rd arrondissement?"

ChatGPT

MCP

Connect Normi directly to ChatGPT via a revocable MCP URL — no raw API token needed. Generate the URL from the dashboard.

Custom AI agents

REST API

Use the REST API to embed live DVF data in your own agent pipelines. OpenAI, Anthropic, Mistral — any LLM that can call HTTP.

Access French Real Estate Data

500 free credits · No credit card required

FAQ

Frequently Asked Questions

Answers to the most common questions about the Normi DVF API.

What is the Normi MCP DVF server?

An MCP server specialized in French real-estate data that exposes 33 tools across DVF, ANIL rents, DPE, BDNB and SIRENE, including location resolution, market statistics, rental yield, purchasing power, short-hold resale detection (detect_flips), comparables, valuations, trends, portfolios and alerts.

How do I install Normi MCP DVF in Claude Desktop?

Add the mcpServers entry with the command npx -y @normi/mcp-dvf and the NORMI_API_KEY token. Generate the free token on the get-token page then restart Claude Desktop.

Can I use the MCP DVF server via HTTP or behind a proxy?

Yes. The streamable-http transport (/mcp) accepts proxies and Authorization Bearer normi_xxx headers. Recommended for Cline, Cursor, VSCode and backends.

How to use HTTP streamable directly (API)?

Make a POST /mcp with method tools/list or tools/call — the server is stateless, no session to manage between calls. Add Authorization Bearer and Accept application/json, text/event-stream (server corrects if needed).

How is DVF data quality ensured?

Automatic deduplication of bulk sales, outlier filtering, intelligent property type mapping and median values calculated before sending to MCP.

What advanced analyses are available with MCP DVF?

Comparable properties, price trends, market heatmaps, address history, neighborhood comparisons and volume activity analysis.

Does Normi provide DPE energy data?

Yes. Normi integrates ADEME energy diagnostics cross-referenced with DVF transactions via cadastral reference. 6 REST endpoints and 3 MCP tools provide energy class distribution (A to G), price premium per DPE class (green premium), thermal sieves F/G and the evolution of the A+B vs F+G price gap over time.