Húsasjá MCP Server for ChatGPT
By Neko Management ehf.
Overview
Húsasjá brings Iceland's housing market into ChatGPT. Ask about homes for sale, what nearby homes have sold for, a specific building or apartment, and what Húsasjá's figures mean. What you can do: - Search homes currently for sale anywhere in Iceland by area, price, size, rooms and features such as a balcony, garage or elevator. Results appear as a row of sale cards with photos, price and size, each linking to the home on husasja.is. - See what is for sale near a given address, sorted by distance. - Look up recent sales and asking prices on a street or in an area, with price per square metre. - Look up a building: construction year, floors, number of units and its approved blueprints. - Find a specific apartment by address and get Húsasjá's estimated market value for it, the same estimate the Húsasjá app shows. - Compare listings side by side, see a listing's details and price history, and get an estimate of how long a home takes to sell. - Get explanations of Húsasjá's concepts, such as the market value estimate, the price index and time on market. Who it is for: people buying, selling or following homes in Iceland, and anyone who wants to understand prices in a neighbourhood. Limitations: - Covers Iceland only. - Read-only. It does not save, change or send anything. - Market value estimates are statistical models built from comparable sales. They are not appraisals by a licensed real estate agent. - Requires signing in with a Húsasjá account.
How to connect
- ChatGPT: Open the Húsasjá listing in the ChatGPT app directory and choose Connect. ChatGPT listing Húsasjá exposes 17 tools here.
Tools
explain_concept
ChatGPTfind_building_by_address
ChatGPTmatch says how the search resolved. single: one building; use its building_id with the other tools. ambiguous: a street and house number matched several buildings (the same address in two towns, say) and one is meant; name the building you would base the answer on (address, construction year), list the others briefly and ask the user which one they mean; never pick one yourself. several: a street or area search, and every building returned is meant; open the answer by saying it covers N buildings ('Miðað við N byggingar…'). none: nothing matched; ask the user to check the address. When truncated is true, more buildings matched than were returned: raise top_k or narrow with postcode or construction year. When unmatched_place is set, the municipality given names no region, municipality or town, so nothing was searched: try the postcode, or the town name in Icelandic.find_current_real_estate_listings
ChatGPTunmatched_place is set, the municipality given names no region, municipality or town, so nothing was searched: use lookup_postcodes to find the area's postcodes. In clients that show apps, the results also appear to the user as a scrollable row of sale cards (photos, price, size, address) they can click through to Húsasjá. There, don't enumerate or tabulate the listings in your reply: write a short summary of what matched (how many, the price and size ranges). In clients without apps, your reply is the user's only view of the results: list the listings that answer their question (address, price, size, rooms, and the Húsasjá link where the result gives one) and say how many matched in all (total_count). Listing ids are for follow-up calls (get_listing_detail, compare_properties), never for the user.find_nearby_listings
ChatGPTtotal_count). Listing ids are for follow-up calls (get_listing_detail, compare_properties), never for the user.find_unit_by_address
ChatGPTmatch says how the address resolved. single: one unit; use its unit_id with the other tools. ambiguous: several units match and one is meant. Clients that show apps show the user these units to pick from, and their pick arrives as their next message: treat it as the answer and resolve the unit it names, without listing the units again. In clients without apps, name the unit you would base the answer on (address, size, rooms), list the other candidates briefly and ask the user which one they mean; never pick one yourself. several: every unit in the building given by building_id, and all of them are meant; open the answer by saying it covers N units in the building ('Miðað við N íbúðir í þessari byggingu…'). none: nothing matched; ask the user to check the address. When truncated is true, more units matched than were returned: narrow the search with postcode, house_number or unit_code. When unit_code_found is false, no unit at the address has the unit code given: tell the user so, then treat the address's units in units as ambiguous.get_building_info
ChatGPTget_expected_days_to_sell
ChatGPTget_listing_detail
ChatGPTget_nearby_amenities
ChatGPTget_sale_cards
ChatGPTsale:123); a bare number is a market-episode id. Called by the view, not the model.get_sales_and_listings_history
ChatGPTget_unit_market_value
ChatGPTget_unit_picker
ChatGPTget_unit_sale_history
ChatGPTlookup_postcodes
ChatGPTplaces is one place the name resolved to. exact false means the name matched by alias, part of the name or approximately, so the place may not be the one meant. A district's or matsvæði's postcodes carry share, how much of the place lies inside each: postcodes don't follow district lines, so a postcode can reach into a neighbouring district. A name can match matsvæði in more than one municipality, and a district split into matsvæði under other names (Reykjavík's Vesturbær, for one) is only partly covered: check the places before using every postcode. Empty places means nothing matched: try the Icelandic name of the town or municipality (e.g. 'Kópavogur', 'Akureyri'), a region such as 'Höfuðborgarsvæðið', or the district's own name.resolve_latest_listing_for_unit
ChatGPTresult is null), use get_unit_market_value for an estimate instead. Needs a unit_id from find_unit_by_address.Capabilities
Example Prompts
Click any prompt to copy it.
Links
Similar Apps
Listed on MCP App Store
Add this badge to your site or README. It links to this page.
HTML
<a href="https://mcp-app-store.keenethics-labs.com/app/h-sasj"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>
Markdown
[](https://mcp-app-store.keenethics-labs.com/app/h-sasj)






