Skip to main content
29M UK properties, one API call Get your API key →
PyPI homedata-mcp v0.2.0 npm homedata-mcp v0.1.0

Property data MCP server for every AI tool — ships in Python and Node

Install the Homedata MCP in any AI coding assistant and let it query 29 million UK properties — EPC, flood risk, planning, schools, transport — without leaving the chat. Pick Python (pip install homedata-mcp) or Node (npx -y homedata-mcp), paste your API key, restart your tool.

Works with Claude Desktop, Claude Code, Cursor, Codex CLI, Windsurf, Cline, Continue.dev and Zed.

Python pip
pip install homedata-mcp
Node npm / npx
npx -y homedata-mcp

Two editions, identical tools

Homedata MCP ships as a Python package and a Node package. Both expose the same MCP tools and the same homedata CLI. Pick whichever runtime fits your AI client and your machine.

Python edition

homedata-mcp v0.2.0

Best when your AI client already embeds Python — Claude Code, locally-managed plugin hosts, scripted automations. Requires Python 3.10+.

pip install homedata-mcp

Installs the homedata-mcp server and the homedata CLI on your PATH.

Node edition

homedata-mcp v0.1.0

Recommended for most AI clients — zero-install via npx, no global PATH dependency. Requires Node 18+.

npm install -g homedata-mcp
# or, no install at all:
npx -y homedata-mcp

Installs the homedata-mcp server and the homedata CLI — same commands, same flags.

How do I install the Homedata MCP?

Drop a small config block into your AI tool, paste your API key, restart. The MCP server runs locally and exposes typed tools to your assistant.

Add to

your AI tool's MCP config file

Pick your tool below for the exact config path. Same JSON shape across Claude, Cursor, Windsurf, Cline and Zed.

Recommended for most AI tools — runs via npx, no global install needed. Node 18+ required.

{
  "mcpServers": {
    "homedata": {
      "command": "npx",
      "args": ["-y", "homedata-mcp"],
      "env": {
        "HOMEDATA_API_KEY": "your_api_key_here"
      }
    }
  }
}
Restart your AI tool after saving. The homedata server appears in its MCP servers list.

Either edition exposes the same MCP tools and the same homedata CLI. Pick Node if you want zero-install via npx; pick Python if your AI client already embeds Python.

1. Install the server

Python 3.10+ with pip, or Node 18+ with npm/npx. Either edition exposes the same MCP tools.

2. Drop in your API key

Sign up and copy your key from the developer dashboard. Paste it into HOMEDATA_API_KEY.

3. Restart your AI tool

Your assistant now has typed property-data tools — ask it about an address.

The homedata CLI — same data, no AI required

Both editions ship a homedata command on your PATH so you can query the API straight from a terminal, shell script or build pipeline. Same auth, same data, same flags as the MCP server.

Install the CLI

Same homedata command in both editions.

Python (pip)

pip install homedata-mcp

Node (npm)

npm install -g homedata-mcp

Or run once, no install

npx -y homedata-mcp homedata --help
Set HOMEDATA_API_KEY in your shell before running any command — same key as the MCP server.

Common commands

Pipe into jq, fzf or feed straight back into your AI tool.

homedata search '10 downing street'

Type-ahead address search — get a UPRN for an address in Great Britain from your shell.

homedata property 100023336956

Full property record by UPRN. Pipe into jq, fzf or your editor.

homedata epc 100023336956 --field current_energy_efficiency

EPC rating, score and recommendations from the official register.

homedata flood-risk 100023336956

Environment Agency flood risk — river, sea, surface and groundwater.

homedata planning --postcode "SW1A 1AA" --days 365

Recent planning applications for a postcode or local authority in England and Wales.

What can my AI ask about UK property?

Typed tools your AI assistant gains the moment the MCP is connected. Each one maps to a documented Homedata endpoint and has its own anchor for deep linking.

Find UK addresses from free text: an address, a postcode or a place name. Returns matching addresses with their UPRN, which every property tool takes.

The cheap first call for a property: which data is available for it, what each add-on costs, and the shortcuts to each tier. Also the quickest way to check whether a UPRN is one we hold.

Address-only record for a property: full address, postcode, coordinates and the standard address identifiers. The cheapest property tier, for address verification, form pre-fill and matching.

The house-hunter view of a property: address, rooms, EPC rating, last sale, construction, dimensions, garden, parking and title basics, in one call.

The full listing view of a property: everything in Base plus council tax band, flood risk, schools, broadband, crime, demographics, solar potential, confirmed sales and planning constraints. The usual starting point.

Everything held on a property in one call: the Core record plus the full council tax charge, full title, environmental risks, deprivation, planning history and area price trends.

Build your own property record: the base record plus only the add-ons you ask for, so you pay for exactly what you use. Call property_discovery first to see which add-ons a property has.

attr_rooms Included

Room counts for a property: bedrooms, bathrooms, habitable rooms and heated rooms.

attr_roof Included

A property's roof: material, shape, and whether solar panels are fitted.

attr_garden Included

Whether a property has a garden, and of what kind.

Whether a property has parking, and of what kind: driveway, off-street, garage and so on.

Improvements recommended by a property's EPC assessment, each with estimated minimum and maximum cost.

Measurements for a property: footprint area in square metres, building height, estimated volume and predicted floor area.

attr_land Included

Plot size in square metres for a property. Houses only: flats and maisonettes have no plot, and those return an error with nothing charged.

How a property was built: construction age band, main construction material, and whether it has a basement.

attr_epc Included

Energy Performance Certificate headline for a property: current and potential efficiency rating, EPC floor area and the date of the last assessment.

risks Included

Environmental risk screening for a property: flood, radon, noise, landfill, coal and other mining, invasive plants and air quality. Ask for one hazard, or for all of them in a single response.

deprivation Included

Index of Multiple Deprivation scores for a postcode, across income, employment, education, health, crime, housing and environment. England only.

How property prices are spread across an outcode area: percentiles, median and transaction counts by property type.

List every registered address at a UK postcode, with the UPRN for each one.

council_tax Included

Council tax band for a property, with the billing authority name and its official code. For the yearly and monthly charge in pounds, use council_tax_full instead.

The full council tax record for a property: band, billing authority and its official code, this year's yearly and monthly charge in pounds, the 1991 valuation bounds behind the band, and the fiscal year.

Land Registry title records for a property: tenure, title number and registered owner where held.

boundaries Included

Search UK administrative areas by name and get their boundary id, for tools that take one.

planning Included

Planning applications near a postcode or coordinates: type, status, description and decision date, with filters for recency, type and status.

crime Included

Recorded crime near a postcode or coordinates, by category and month, from Police UK.

solar Included

Solar potential for a property: usable roof area, estimated yearly generation, savings, payback period and carbon saved.

One-call summary of a postcode: deprivation, crime, average property price, nearby schools, transport and broadband. Cheaper than calling those tools separately.

Capital growth for an outcode area: annual growth rate, returns over one, three, five and ten years, and a historical price index, from Land Registry sold prices.

schools Included

Schools near a postcode, with Ofsted rating, phase, pupil numbers and distance, from the Department for Education register. England only.

broadband Included

Broadband availability at a postcode: average and maximum download and upload speeds, superfast, ultrafast, gigabit and full-fibre coverage, and how many premises are covered. From Ofcom Connected Nations.

Every amenity group near a property in one response: food, education, healthcare, financial, civic, worship, culture, convenience, green spaces, transport and shops.

Places to eat and drink near a property: cafes, restaurants, pubs, bars, takeaways.

Education near a property: schools, nurseries, colleges, universities, childcare and libraries. For Ofsted ratings and pupil numbers use schools.

Health services near a property as mapped locally: doctors, dentists, clinics, hospitals, pharmacies and care homes. For regulator-registered records use healthcare_all and its per-type tools.

Banks, cash machines and money services near a property.

Civic places near a property: post offices, town halls, courthouses, fire and police stations, community centres.

Places of worship near a property.

Culture and leisure near a property: theatres, cinemas, music venues, museums, galleries and attractions.

Everyday conveniences near a property: public toilets, charging points, parcel lockers and similar.

Green space near a property: parks, playgrounds, gardens, nature reserves, commons and sports pitches.

Transport stops near a property: bus stops, railway stations, tram stops and ferry terminals.

Shops near a property, from supermarkets to specialist retailers.

Petrol stations and EV charging points near a property, each tagged with which it is.

Petrol stations near a property.

EV charging points near a property.

Health services near a property: regulator-registered GPs, dentists and hospitals, plus pharmacies. Registered entries carry name, address, postcode, region, distance and a link to the official register record. Registered records cover England.

Registered GP practices near a property, with name, address, postcode, region, distance and a link to the official register record. England only.

Registered dental practices near a property, with name, address, postcode, region, distance and a link to the official register record. England only.

Pharmacies near a property, with name, address, phone and website where known.

Registered hospitals near a property, with name, address, postcode, region, distance and a link to the official register record. England only.

Census 2021 profile for the area around a postcode: population, tenure, age bands, ethnicity, occupation, household size and car ownership, plus deprivation where available.

Listed buildings within a radius of a postcode: Grade I, II* and II entries with name, location, listing date and a link to the official record.

Work out Stamp Duty Land Tax for England and Northern Ireland: total tax, effective rate and a band-by-band breakdown, for a main residence, a first-time buyer or an additional property.

Work out a mortgage from price, deposit, rate and term: monthly payment, total repayment, total interest, and loan-to-value and loan-to-income ratios.

Tools call the same endpoints documented on /docs/endpoints. Property data sourced from Home.co.uk, Land Registry, EPC register, Environment Agency, ONS, Ofsted and other authoritative UK sources.

Which AI tools support MCP?

MCP is an open protocol from Anthropic. Adoption is broad and growing — Homedata works with the major coding-focused clients today.

Claude Desktop

Setup guide →

Anthropic's official desktop app for Claude.

~/Library/Application Support/Claude/claude_desktop_config.json

Claude Code

Setup guide →

Anthropic's terminal-native coding agent.

~/.config/claude-code/mcp.json

AI-first code editor built on VS Code.

~/.cursor/mcp.json

Codex CLI

Setup guide →

OpenAI's terminal-native coding agent.

~/.codex/config.toml

Windsurf

Setup guide →

Codeium's agentic IDE built for flow.

~/.codeium/windsurf/mcp_config.json

VS Code extension by the Cline team — autonomous coding inside your editor.

VS Code → Cline → MCP Servers → Edit MCP Settings

Continue.dev

Setup guide →

Open-source AI assistant for VS Code and JetBrains.

~/.continue/config.yaml

High-performance, multiplayer code editor.

~/.config/zed/settings.json

Frequently asked questions about the Homedata MCP

Common questions about installing, supported AI tools and underlying data sources.

What is the Model Context Protocol (MCP)?
MCP is an open standard from Anthropic that lets AI assistants like Claude, Cursor, Codex and Windsurf talk to external tools and data sources. When you install the Homedata MCP, your AI gains typed tools for querying UK property data — addresses, EPC, flood risk, planning, schools, transport and more — directly from your normal chat or coding workflow.
Which AI tools support the Homedata MCP?
Claude Desktop, Claude Code, Cursor, Codex CLI, Windsurf, Cline (VS Code), Continue.dev and Zed all ship with first-class MCP support. The Homedata MCP ships as both a Python package (pip install homedata-mcp) and a Node package (npm install -g homedata-mcp / npx -y homedata-mcp) so it slots into every one of them — pick the dedicated subpage above for setup instructions tailored to your tool.
How does the Homedata MCP work?
When you install homedata-mcp and configure your AI client with your Homedata API key, the assistant gains a set of typed property-data tools. Ask it a question about a UK address and the assistant decides which tools to invoke, makes the request via the MCP server running locally on your machine, and uses the structured response to answer you. The MCP server runs locally — your AI vendor never sees the property data.
Should I install the Python or Node edition of homedata-mcp?
They expose the same MCP tools and the same homedata CLI — pick whichever runtime is easier on your machine. Node is recommended for most AI clients (Claude Desktop, Cursor, Codex, Windsurf, Cline, Continue.dev, Zed) because npx lets you run it with no global install. Python is recommended for Claude Code and any AI tool that already embeds Python.
What UK property data is available through MCP?
The MCP exposes typed tools for 29 million UK properties — address search, full property records, EPC ratings, flood and environmental risk, planning applications, crime statistics, Ofsted school data, transport links and aggregated postcode profiles. Sales history and live listings are included too — every plan reaches every tool.
How do I install the Homedata MCP?
Pick an edition: `pip install homedata-mcp` for Python, or `npm install -g homedata-mcp` (or `npx -y homedata-mcp`) for Node. Then drop a small JSON or TOML block into your AI tool's config file with your Homedata API key. Each AI client has its own config path — pick the dedicated subpage above for the exact snippet and location for your tool.
How does the homedata CLI differ from the MCP server?
The MCP server is for AI assistants — it speaks the Model Context Protocol so Claude, Cursor and friends can call typed tools. The homedata CLI is for humans — it ships in the same package and gives you `homedata property`, `homedata epc`, `homedata flood-risk`, `homedata planning` and `homedata search` commands you can run directly in a terminal, pipe into jq, or use from shell scripts.
Is the Homedata MCP open source?
Yes — both editions are MIT-licensed. The Python source lives at github.com/wehomemove/homedata-mcp and the Node source at github.com/wehomemove/homedata-mcp-node. You can read every tool definition, fork either repo, contribute new tools and pin a specific version if you need stability across deploys.
Why is UK property data hard?
UK property data is fragmented across dozens of separate registers — Land Registry for sales, the EPC register for energy, the Environment Agency for flood, local authorities for planning, ONS for census, Ofsted for schools, Home.co.uk for live listings. Homedata aggregates them into a single typed API. The MCP wraps that API so an AI assistant can query any of those sources without the developer learning each one.
How do I get help or report a bug?
Open an issue on the public GitHub repository (homedata-mcp or homedata-mcp-node), or email support@homedata.co.uk. Both editions are MIT-licensed and PRs are welcome.

Plug 29 million UK properties into your AI assistant

Works with every major AI coding tool. Install in under sixty seconds — Python or Node.

Latest releases: PyPI v0.2.0 · npm v0.1.0