Install guide / n8n
Connect Normi to an n8n AI agent
In n8n, the MCP Client Tool node gives an AI Agent access to the Normi tools: DVF sales, DPE energy ratings and the BDNB, inside your automated workflows (market monitoring, lead enrichment, reports).
https://mcp.normi.fr/mcpSettings checked on 10 October 2026 against the n8n documentation and the MCP Client Tool node source code (node version 1.4). n8n MCP Client Tool documentation
Before you start
- A Normi account: sign in with Google, 500 free credits a month.
- For the API-key connection or the STDIO package: a
normi_…key, created in Dashboard → Tokens. It is shown only once. - For the STDIO package: Node.js 18 or later (
npx).
Remote connection (HTTP Streamable)Recommended
- In your workflow, add an AI Agent node and connect a chat model to it.
- On the agent's Tool input, add MCP Client Tool.
- Fill in the Endpoint and the HTTP Streamable transport.
- Authentication Bearer Auth: create a credential with your
normi_…key as the token. - Choose which tools the agent can see, then run the workflow.
Endpoint: https://mcp.normi.fr/mcp Server Transport: HTTP Streamable Authentication: Bearer Auth Credential → Bearer Token: normi_YOUR_TOKEN Tools to Include: Selected → analyze_dpe_price_premium, get_market_overview, resolve_location
The node also offers MCP OAuth2. An API key is still the simplest choice for a workflow that runs unattended: it doesn't depend on any browser session.
Scoping the agent
A short system message avoids unnecessary calls. Normi requires a location filter on every request: asking for resolve_location first avoids errors on municipality names.
You are a real estate analyst. For any question about a place, call resolve_location first to get the canonical filters, then the relevant Normi tool. Always state how many sales were used.
STDIO package
n8n's built-in node doesn't start STDIO servers: use the remote connection above. To call Normi without an agent, the HTTP Request node also works with the REST API (reference in French), which exposes the same data.
A first prompt to try
The agent calls detect_flips (10 credits). An excerpt of the real response, captured on 10 October 2026 against the 2014–2025 DVF data:
{
"commune": "RENNES",
"periode": { "mois": 24, "date_debut": "2023-12-31", "date_fin": "2025-12-31" },
"type_local": "Maison",
"transactions_revente": 938,
"flips_detectes": 11,
"taux_de_flip": 1.17,
"marge_brute_mediane": -6040,
"marge_mediane_pct": -1.44,
"duree_detention_mediane_jours": 651
}Over 24 months, 11 of the 938 house resales in Rennes (1.17%) are quick resales, with a median gross margin of -€6,040 (-1.4%). In a workflow, connect the agent's output to an email or a Google Sheets row for a monthly report. The response fields are in French because the API's field names are.