Airbnb MCP: filtered listing search and details for agents

Booking a stay mostly means fiddling with filters: dates, adults and kids, pets, a nightly budget, then opening listing after listing to see whether there’s a washer. This MCP server hands Airbnb search and listing details to an agent, so you state the constraints once and get back linked results. It isn’t an Airbnb product, and it had about 550 stars as of 2026-10-06.

What it does

• airbnb_search: search by location with check-in and check-out dates, adults, children, infants and pets, a price range, and room type (entire home, private room, shared room, hotel room), with pagination.
• airbnb_listing_details: amenities, house rules, highlights and coordinates for a listing ID. Amenities are grouped by category, and anything Airbnb shows struck through lands under “Not included”, so the agent won’t claim a listing has it.
• Better non-US results: the server geocodes the place name through Photon or Nominatim first, because Airbnb’s own geocoder sends queries like “Paris, France” to the wrong place. Pass a Google Maps Place ID to skip that step.
• Polite by default: honors robots.txt, times out stuck requests, and lets you switch off third-party geocoding.

Who it’s for

• Families planning a trip who want an agent to make the first pass on budget and headcount.
• Developers building travel agents who need a ready source of listing data.

Setup

Needs Node.js 18+. No account or API key. For Cursor and similar clients:

{
  "mcpServers": {
    "airbnb": {
      "command": "npx",
      "args": ["-y", "@openbnb/mcp-server-airbnb"]
    }
  }
}

Claude Desktop users can instead download the .mcpb bundle from the latest release and open it.

Our take

Among travel servers this one has the tidiest tool design: two tools cover searching and reading, and the README is upfront about every outside service it calls. Under the hood it parses Airbnb’s web pages, which brings two caveats. A site redesign can break it, as the project’s own tests acknowledge, and heavy scraping risks breaching Airbnb’s terms. There’s a switch to ignore robots.txt, which the README says to keep for testing. It searches and reads only; it can’t book. MIT licensed.