THIS IS A TOY IMPLEMENTATION - USE AT YOUR OWN RISK

nosy-neighbour

Look up Danish property records from tinglysning.dk via a map-based browser UI or a command-line tool.

Given any freeform Danish address you get:

• Owners (ejere) with ownership share
• Valuation (vurdering) — property value, land value, municipality
• Estimated equity (friværdi) — valuation minus total registered mortgage principals
• Mortgages and liens (hæftelser) — including estimated loan type (F-kort / F1 / F3 / F5) for variable-rate realkreditlån
• Easements (servitutter)

Loan type estimation works by matching the registered coupon rate against Nationalbanken rate statistics (DST table DNRNURI). When an ISIN is known, the type is confirmed definitively via ESMA FIRDS.

---

Running locally

Requirements: Python 3.10+

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# Web UI (http://localhost:8000)
python server.py

# CLI
python nosy_neighbour.py "Stålhøjen 24, 8240 Risskov"

---

Running with Docker

# Build
docker build -t nosy-neighbour .

# Run (web UI on http://localhost:8000)
docker run -p 8000:8000 nosy-neighbour

Production deployment (Docker Compose + Caddy)

A compose.yml is included that runs the app behind a Caddy reverse proxy with automatic HTTPS via DuckDNS.

cp .env.example .env
# edit .env with your values
docker compose up -d

Caddy listens on port 18000 and proxies to the app. Set DUCKDNS_DOMAIN to your own DuckDNS subdomain (e.g. yourname.duckdns.org).

---

Web UI

Open http://localhost:8000 in a browser.

• Search by typing an address in the sidebar — autocomplete is powered by DAWA.
• Click the map to look up the property at that location (reverse geocoding via Dataforsyningen).
• Up to 10 addresses can be pinned simultaneously. Each gets a numbered marker; click a marker to scroll to its details in the sidebar.
• Remove any address with the × button in its header.

---

API

Endpoint	Parameters	Description
GET /api/autocomplete	q	DAWA address autocomplete
GET /api/reverse	lat, lng	Reverse geocode a map click to an address
GET /api/lookup	q	Full property lookup by freeform address

Example

curl "http://localhost:8000/api/lookup?q=Stålhøjen+24+Risskov"

{
  "adresse": "Stålhøjen 24, 8240 Risskov",
  "ejendomstype": "Ejerlejlighed",
  "matrikler": [{ "matrikelnummer": "...", "landsejerlavnavn": "..." }],
  "vurdering": {
    "vurderingsdato": "2022-01-01",
    "ejendomsvaerdi": 2650000,
    "grundvaerdi": 312500,
    "kommune": "Aarhus"
  },
  "ejere": [{ "navn": "...", "andel": "1/1" }],
  "haeftelser": [
    {
      "prioritet": "1",
      "haeftelsestype": "Realkreditpantebrev",
      "hovedstol": "1.716.000 kr.",
      "rente": "1.5",
      "fastvariabel": "variabel",
      "kreditorer": ["Totalkredit A/S"],
      "loan_type_info": {
        "source": "estimated",   // or "esma_firds" when confirmed via ISIN
        "loan_type": "F3",
        "uncertain": false,      // true when the rate falls close to a boundary
        "close_to": [],          // nearby loan types when uncertain is true
        "candidates": [...]      // all rate-matched candidates considered
      }
    }
  ],
  "servitutter": [...]
}

---

MCP server

The service exposes an MCP server at POST /mcp using the Streamable HTTP transport. Connect any MCP-compatible client (Claude Desktop, Claude Code, etc.) to http://localhost:8000/mcp.

Tool: lookup_property

Parameter	Type	Description
address	string	Freeform Danish address

Returns the same JSON payload as GET /api/lookup.

Claude Desktop config

{
  "mcpServers": {
    "nosy-neighbour": {
      "url": "http://localhost:8000/mcp"
    }
  }
}

Claude Code

claude mcp add --transport http nosy-neighbour http://localhost:8000/mcp

---

CLI

usage: nosy_neighbour.py [-h] [--isin PRIORITY:ISIN] address [address ...]

positional arguments:
  address              Freeform address, e.g. "Molsvej 38 6950 Ringkøbing"

options:
  --isin PRIORITY:ISIN  ISIN for a specific mortgage, as priority:ISIN
                        (e.g. --isin 1:DK0004632486). Can be repeated.

When an ISIN is supplied for a mortgage, the loan type is resolved definitively from ESMA FIRDS instead of estimated from rate statistics.