Skip to main content
Glama
AgentIA1984-cmd

MCP Immobilier France (DVF)

by AgentIA1984-cmd
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Remote

france-property-mcp-server

MCP server for French real-estate intelligence, built entirely on open French government data:

  • Base Adresse Nationale (BAN) — address geocoding → coordinates + INSEE code.

  • DVF (Demandes de Valeurs Foncières) — recorded property sale transactions since 2010.

It turns raw open data into agent-usable tools: geocoding, sale history, comparable-based price estimation, and commune market statistics with a year-over-year trend. This is the sellable v1 of the AgentIA Off-Market Property Radar MCP; the pure off-market lead layer (successions, saisies, ventes notariales) is a premium roadmap item — see Roadmap.

Tools

Tool

What it does

Key inputs

fr_property_geocode

Resolve a French address to lat/lon + INSEE code

address, limit

fr_property_transactions

List DVF sales near an address / in a commune

address or citycode, radius_m, property_type, since_year, min_rooms, max_rooms, limit

fr_property_price_estimate

Estimate a property's value from comparables (median €/m² × surface, p25–p75 range, confidence)

address, surface_m2, property_type, radius_m, years_back

fr_property_market_stats

Commune median/p25/p75 €/m², volume, YoY trend

address or citycode, property_type, years_back

All tools are read-only, support response_format: "markdown" | "json", and return both text and structuredContent.

Related MCP server: immo-mcp

Quickstart

npm install
npm run build

# Local (stdio) — for Claude Desktop, Cursor, etc.
npm start

# Remote (HTTP) — for hosted deployment
TRANSPORT=http PORT=3000 npm start

Connect to a local MCP client (stdio)

{
  "mcpServers": {
    "france-property": {
      "command": "node",
      "args": ["/absolute/path/to/france-property-mcp-server/dist/index.js"]
    }
  }
}

Test the HTTP transport

curl -s -X POST http://localhost:3000/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"c","version":"1.0.0"}}}'

Note on networking: the tools call api-adresse.data.gouv.fr (BAN) and the DVF API. These are public and work from any normal host. In restricted sandboxes with egress filtering you may see Error: upstream API returned status 403 — that is the sandbox, not the server.

Configuration (env)

Variable

Default

Purpose

TRANSPORT

stdio

stdio or http

PORT

3000

HTTP port

BAN_API_BASE

https://api-adresse.data.gouv.fr

Geocoding endpoint

DVF_API_BASE

https://api.cquest.org/dvf

DVF transactions endpoint

SERVER_API_KEY

(empty)

If set, HTTP clients must send header x-api-key

Deploying it to sell

An MCP is sold as a hosted service whose tool-calls are billed — not as a downloadable file.

  1. MCPize (recommended) — publish the server; the platform handles hosting, SSL, Stripe checkout, license keys, analytics (≈85/15 revenue share). Set TRANSPORT=http and, if you gate access yourself, SERVER_API_KEY.

  2. Apify — first-class MCP hosting with usage events (≈80% net). Wrap dist/index.js as the actor entry.

  3. Visibility (no payment): list on Glama, mcp.so, PulseMCP, Smithery for discovery/SEO.

A /health endpoint and an optional x-api-key gate are included for hosting platforms.

Suggested pricing

  • Freemium: geocode + a few transaction lookups free.

  • Pro subscription: €19–39/mo for unlimited price_estimate + market_stats (the value tools).

  • Per-call: for API/agent buyers who prefer usage-based billing.

⚠️ Production reliability — data sourcing (the key risk)

The default DVF_API_BASE (api.cquest.org/dvf) is a community proof-of-concept with no availability guarantee. Before selling, do one of:

  • Self-host the micro-API from https://github.com/cquest/dvf_as_api and point DVF_API_BASE at it (cheap, removes the dependency), or

  • Point DVF_API_BASE at a DVF+ provider (Cerema/SOGEFI "API données foncières" on data.gouv.fr).

The DVF client normalizes several field-name variants, so most providers work by swapping the base URL. Data is under the Licence Ouverte / Etalab (attribution required).

Roadmap (premium "off-market" layer)

  • Ventes notariales / successions / saisies feeds (higher willingness-to-pay).

  • Cadastre parcel lookup (apicarto.ign.fr) and DPE energy label (ADEME) enrichment.

  • Rental-yield estimation (combine sale €/m² with rent references).

  • Alerting webhook (new matching sale in a saved zone).

Disclaimer

fr_property_price_estimate returns a statistical estimate from public comparable sales — it is not a certified valuation (expertise immobilière). Verify before any transaction decision.

Available Tools

4 tools
fr_property_geocodeGeocode French addressA
Read-onlyIdempotent

Resolve a free-text French address to geographic coordinates and its INSEE commune code using the Base Adresse Nationale (BAN).

Use this to obtain the latitude/longitude and 'citycode' (INSEE code) needed by the other tools, or to validate/normalize an address.

Args:

  • address (string): free-text French address

  • limit (number): max candidate matches, 1-20 (default 5)

  • response_format ('markdown' | 'json'): output format (default markdown)

Returns: ranked candidates with label, lat, lon, citycode (INSEE), postcode, city, score (0-1), type.

ParametersJSON Schema
NameRequiredDescriptionDefault
limitNoMaximum number of candidate matches to return (1-20)
addressYesFree-text French address, e.g. '12 rue de Rivoli, Paris'
response_formatNoOutput format: 'markdown' (human) or 'json' (machine)markdown

TDQS

A4.5/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnlyHint, idempotentHint, and destructiveHint as true/false, so the safety profile is established. The description adds value by stating the underlying data source (BAN), the ranking mechanism (score 0-1), and the specific output fields, which go beyond annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is well-structured with a clear first sentence stating purpose, followed by usage guidance, parameter list, and return description. No extraneous information; every sentence adds value. It is concise yet thorough.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool has 3 parameters (all documented) and no output schema, the description compensates by listing the returned fields (label, lat, lon, citycode, postcode, city, score, type) and their meaning. It also provides context about the BAN source and the tool's role in the suite, making it fully informative for an AI agent.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Input schema has full parameter description coverage (100%), so baseline is 3. However, the description adds context: it explains that 'address' is free-text, 'limit' is maximum candidates with default 5, and 'response_format' controls output type. This enriches understanding beyond the schema's minimal descriptions.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the verb 'resolve' and the resource 'free-text French address' to obtain geographic coordinates and INSEE code. It distinguishes from sibling tools by noting the output is needed by other tools, making its purpose unambiguous.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description explicitly advises when to use the tool: to obtain latitude/longitude and citycode for other tools, or to validate/normalize an address. It does not mention when not to use or direct alternatives, but the context is clear given sibling tools serve different purposes.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

fr_property_market_statsFrench commune market stats (DVF)A
Read-onlyIdempotent

Compute commune-level real-estate market statistics from DVF: median/p25/p75 EUR/m², sales volume, and year-over-year trend for a property type.

Provide EITHER 'address' (its commune is used) OR 'citycode' (INSEE).

Args:

  • address (string, optional): address inside the commune

  • citycode (string, optional): INSEE commune code

  • property_type (string): 'Appartement' (default) or 'Maison', etc.

  • years_back (number): lookback window (default 4)

  • response_format ('markdown' | 'json')

Returns: salesCount, medianPricePerM2, p25/p75 EUR/m², medianPrice, and yoyPricePerM2Pct (last full year vs prior year, null if insufficient data).

ParametersJSON Schema
NameRequiredDescriptionDefault
addressNoAddress inside the commune to analyse (or use 'citycode').
citycodeNoINSEE commune code (5 chars). Alternative to 'address'.
years_backNoLookback window in years for stats (default 4).
property_typeNoProperty type to analyse, e.g. 'Appartement' or 'Maison'.Appartement
response_formatNoOutput format: 'markdown' (human) or 'json' (machine).markdown

TDQS

A4.6/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations declare readOnlyHint, openWorldHint, idempotentHint as true, and the description adds that yoyPricePerM2Pct returns null if insufficient data, disclosing edge-case behavior. No contradictions.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Two concise paragraphs, front-loaded with core purpose, then parameter guidance, then return values. Every sentence adds value with no redundancy.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

The description covers parameters and return values well, but lacks notes on data freshness, error handling, or performance. Still highly adequate given good annotations.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, but the description adds clarity by summarizing the exclusivity of address/citycode and listing return fields not in the schema. It goes beyond the baseline 3.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states it computes commune-level real-estate market statistics from DVF, specifying exact metrics like median/p25/p75 EUR/m² and sales volume. It distinguishes from sibling tools (geocode, price estimate, transactions) by focusing on aggregative stats.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It explicitly says to provide either 'address' or 'citycode', guiding parameter selection. It does not directly contrast with siblings for when to use this tool vs alternatives, but the purpose is clear enough.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

fr_property_price_estimateEstimate French property value (DVF comparables)A
Read-onlyIdempotent

Estimate the market value of a French property using comparable DVF sales (same type, similar surface, nearby, recent).

Method: geocode the address, select comparable sales within 'radius_m' of the same 'property_type' with a built surface within ±35% and sold within 'years_back' years, then apply the median EUR/m² to the property's surface. Returns a point estimate and a p25-p75 range with a confidence level based on the number of comparables.

Args:

  • address (string): property address

  • surface_m2 (number): built surface in m²

  • property_type (string): 'Appartement' (default) or 'Maison', etc.

  • radius_m (number): comparable radius in metres (default 500)

  • years_back (number): lookback window in years (default 5)

  • response_format ('markdown' | 'json'): output format

Returns: estimate (EUR), low/high range (EUR), median/p25/p75 EUR/m², comps count, confidence ('high'|'medium'|'low'|'insufficient'). This is a statistical estimate, not a certified valuation.

ParametersJSON Schema
NameRequiredDescriptionDefault
addressYesAddress of the property to estimate.
radius_mNoRadius for comparable selection, metres (50-5000).
surface_m2YesBuilt surface of the property in square metres.
years_backNoHow many years of past sales to consider (default 5).
property_typeNoProperty type to compare against, e.g. 'Appartement' or 'Maison'.Appartement
response_formatNoOutput format: 'markdown' (human) or 'json' (machine).markdown

TDQS

A4.5/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already indicate read-only, idempotent, and non-destructive behavior. The description adds valuable context: it details the estimation methodology (geocode, ±35% surface, median EUR/m²), warns it's a statistical estimate not certified, and explains confidence levels. This goes beyond annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is well-structured: purpose sentence, method paragraph, args list, returns. It is front-loaded and every sentence adds value, including the caveat about statistical estimate. No filler.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Despite no output schema, the description thoroughly explains return values (estimate, range, comps count, confidence) and the method. It covers the algorithmic details, default values, and limitations (not certified valuation), making it fully actionable for an agent.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100% (6/6 parameters described in schema). The description adds methodological context (e.g., radius_m for comparable selection, years_back lookback) but does not significantly enhance individual parameter meanings beyond what the schema already provides. Baseline 3 is appropriate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states it estimates the market value of a French property using DVF comparables, specifying the verb (estimate) and resource (French property). It distinguishes from sibling tools like geocode or transactions by focusing on valuation.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description explains the method and what the tool does, but does not explicitly state when to use it over alternatives like fr_property_market_stats or fr_property_transactions. However, the purpose is clear enough for an agent to infer usage.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

fr_property_transactionsList French property sales (DVF)A
Read-onlyIdempotent

List recorded real-estate sale transactions from the French DVF open dataset, near an address or within a commune.

Provide EITHER 'address' (recommended — enables radius search) OR 'citycode' (INSEE, whole commune).

Args:

  • address (string, optional): address to search around

  • citycode (string, optional): INSEE commune code (5 chars)

  • radius_m (number): radius around the address in metres (default 500)

  • property_type (string, optional): e.g. 'Maison', 'Appartement'

  • since_year (number, optional): earliest sale year (>= 2010)

  • min_rooms / max_rooms (number, optional): room-count filter

  • limit (number): max results, 1-200 (default 50)

  • response_format ('markdown' | 'json'): output format (default markdown)

Returns: sales with date, price (EUR), property type, surface (m2), rooms, price/m2, address, commune. Sorted most-recent first.

ParametersJSON Schema
NameRequiredDescriptionDefault
limitNoMaximum number of transactions to return (1-200).
addressNoAddress to search around (recommended). Either 'address' or 'citycode' is required.
citycodeNoINSEE commune code (5 chars). Alternative to 'address'.
radius_mNoSearch radius in metres around the address (50-5000).
max_roomsNoMaximum number of main rooms.
min_roomsNoMinimum number of main rooms.
since_yearNoKeep only sales on/after January 1 of this year (DVF starts 2010).
property_typeNoFilter by property type substring, e.g. 'Maison' or 'Appartement'.
response_formatNoOutput format: 'markdown' (human) or 'json' (machine).markdown

TDQS

A4.4/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already indicate readOnlyHint, openWorldHint, idempotentHint, and destructiveHint=false. The description adds useful behavioral details such as the return format (sales with fields, sorted most-recent first) and data source (DVF starts 2010), which go beyond annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is front-loaded with purpose and key choice, followed by a clean argument list. It is concise for 9 parameters, with no extraneous information. Minor improvement could be a more structured list, but overall efficient.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's complexity (9 parameters, no output schema), the description covers all necessary aspects: data source, required conditions, filtering options, return fields, sorting, and format. Annotations further cover safety and idempotency, making the description complete for an agent to use effectively.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

With 100% schema coverage, the description still adds value by clarifying the mutual exclusivity of 'address' and 'citycode', emphasizing the default radius, and grouping parameters into logical categories (e.g., filtering by rooms, year). This provides meaning beyond the schema alone.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool lists real-estate sale transactions from the French DVF dataset, with options for address or commune. This specific verb+resource+scope distinguishes it from sibling tools like geocoding, market stats, or price estimates.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides clear context on when to use the tool, advising to provide either 'address' or 'citycode'. It recommends 'address' for radius search. However, it does not explicitly mention when not to use it or directly reference sibling tools for alternatives.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 4 tool updatesv1.0.0
    • First observedfr_property_geocode
    • First observedfr_property_market_stats
    • First observedfr_property_price_estimate
    • First observedfr_property_transactions
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A4.6/5.0

Scored across 4 tools

Disambiguation5/5

Each tool has a distinct purpose: geocoding addresses, computing commune-level market statistics, estimating property value via comparables, and listing recent transactions. There is no overlap or ambiguity.

Naming Consistency5/5

All tool names follow a consistent pattern: 'fr_property_' followed by a descriptive noun or verb (geocode, market_stats, price_estimate, transactions). The naming is uniform and predictable.

Tool Count5/5

With 4 tools, the set is well-scoped for the domain of French property analysis. Each tool serves a necessary function without redundancy, and the count is within the typical ideal range of 3-15.

Completeness4/5

The tools cover the core workflow of geocoding, market statistics, price estimation, and transaction listing. Minor gaps exist (e.g., no direct property search or historical trends), but the surface is largely complete for the stated purpose.

Maintenance

ActivitySlowing
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers