Every market speaks the same shape — you learn the API once and get 16 countries and cities. Location format auto-adapts per market (postcode, ZIP, area name).
Recent sold prices within a radius of any location, filtered by property type, bedrooms and date range.
Median, average and percentile prices for an area, broken down by property type.
Monthly price series with year-over-year change, for charts and market-timing signals.
// what's sold near this flat lately? const res = await fetch( "https://api.nwc-advisory.com/v1/comps?" + new URLSearchParams({ market: "uk", postcode: "E5", property_type: "F", radius_miles: "0.5", months: "24", limit: "25" }), { headers: { "X-RapidAPI-Key": process.env.KEY } } ); const { results, _meta } = await res.json();
import requests r = requests.get( "https://api.nwc-advisory.com/v1/comps", params={ "market": "nyc", "zip_code": "10001", "months": 24, "limit": 25, }, headers={"X-RapidAPI-Key": KEY}, ) comps = r.json()["results"]
curl "https://api.nwc-advisory.com/v1/stats?market=fr&code_postal=75001" \ -H "X-RapidAPI-Key: YOUR_KEY"
Drop the open-source MCP server into Claude, ChatGPT or any agent framework and it can pull comparables and area stats on its own — no glue code, no wrapper to maintain.
search_property_comps find recent sales near a locationget_area_stats median & percentile prices for an arealist_markets discover all 16 markets and their filters# pip install mcp httpx, clone the repo, then: { "mcpServers": { "property-comps": { "command": "python", "args": ["path/to/mcp_server.py"] } } } # free, no key. then just ask your agent: # "what have 2-beds sold for near E5 this year?"
A production app runs on it today — and because it's a clean HTTP API with an open-source MCP server, adding it is an afternoon, not a project.
A production application pulls UK comparable sales through the API to value homes for its users — real daily traffic on the paid tier.
Fork it, wire it into your agent, or run it as-is. Three tools, no boilerplate — the fastest way to give an LLM property data.
github.com/Tianning-lab/property-comps-mcp-server ↗Generate a typed client in any language, or import straight into Postman — the whole surface is documented and machine-readable.
openapi.json ↗Each market maps to its own official registry. The location parameter adapts; everything else stays identical.
Billed through RapidAPI, so it drops into whatever you're already building. No contract, cancel anytime.
Grab a free key, make your first call in a minute, and only pay when your app takes off.