Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@HomeScout SF Show me homes in San Francisco under $2M in the Mission."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An mcp-use MCP App with a Zillow-style split view: San Francisco listing
cards on one side and a live map with price pins on the other.
The model opens the view once with search-homes, then refines it in place
through tools the view registers itself.
All listings are fictional. Map tiles come from Esri's keyless Canvas basemaps (light or dark gray, following the host theme). No listing API, API key, or paid service is involved.
Open the chat demo and ask for homes in San Francisco. Select Fullscreen to see the listing cards. The demo chat uses Manufact's managed model, which cannot call view tools, so run the example locally to try follow-up refinements.
MCP endpoint: https://manufact-property-search-example.run.mcp-use.com/mcp
Related MCP server: Airbnb Search & Listings
Requires Node.js 22.22.2 or later.
git clone https://github.com/manufacts/mcp-use-property-search-example.git
cd mcp-use-property-search-example
npm ci
npm run devOpen the Inspector URL printed by the CLI, select Chat, and configure a model provider with your own API key. The Inspector forwards view tools to the model only in this mode; the managed Manufact model cannot call them. Try these prompts in order:
"Show me homes in San Francisco."
"Now search the Mission, under $2M."
"Remove the two most expensive homes."
"Open the cheapest one on the map and save it."
"Zoom in one step."
The first prompt calls search-homes and opens the view. Every later prompt
calls a view tool, so the open map updates without rendering a second view.
Click Fullscreen in the view to switch display modes.
search-homes returns the matching listing IDs plus the whole staged catalog in
structuredContent, so the view can filter any neighborhood locally. After it renders, the view registers tools
with useViewTool that filter, select, and move the map locally.
Tool | Called by | Purpose |
| Model | Open the view with an initial search |
| View (app-only) | Load extra facts when a card or pin is selected |
| Model, via the view | Change area, price, beds, baths, home type, or sort in place |
| Model, via the view | Hide homes from the cards and map |
| Model, via the view | Fly to a home and open its detail card |
| Model, via the view | Save or unsave homes |
| Model, via the view | Fit every visible home in frame |
| Model, via the view | Zoom in or out |
| Model, via the view | Pan north, south, east, or west |
The view also reports its current area, filters, and sort to the model with
ModelContext, and sends follow-up messages with useSendFollowUp.
The catalog covers Pacific Heights, Marina, Russian Hill, Nob Hill, Hayes Valley, SoMa, Mission District, Noe Valley, Potrero Hill, and Bernal Heights.
src/index.ts: the staged catalog, search-homes, and get-listing-details.
views/property-search/view.tsx: the view, its view tools, and model context.
views/property-search/map.tsx: the Leaflet map, Esri tiles, pins, and camera controls.
views/property-search/cards.tsx: result cards and the detail panel.
npm run typecheck
npm run buildBuild with npm ci && npm run build and start with npm start. The MCP
endpoint is /mcp. No environment variables are required.
Source: mcp-use property search example.
This server cannot be deployed
AI-native real estate discovery with structured property search and market intelligence.
Pull property listings, prices, and details from real-estate sites as structured JSON.
Searchable directory of Airbnb listings — discover properties and retrieve direct Airbnb links.
Redfin listings, sale-comps, and neighborhood market data via natural-language queries.