Airbnb MCP Server

Destination and stay preferences drive airbnb_search, which lets an AI agent find Airbnb listings by location, travel dates, guest composition, nightly price limits and accommodation type. Supported types include entire homes, private rooms, shared rooms and hotel rooms. Results contain pricing and listing links, while a pagination cursor lets the client request additional results. After selecting a listing, airbnb_listing_details accepts its ID to retrieve amenities, coordinates, descriptions, highlights and house policies. Date and guest parameters can also accompany that details request.

Location handling matters for international searches: without a Google Maps Place ID, the server uses Photon for geocoding, with Nominatim as a fallback. Only the supplied location text is sent to these services. Providing a Place ID bypasses that step; geocoding can also be disabled globally, though Airbnb's own location resolution may misidentify non-US destinations. The project uses Node.js and communicates with MCP clients over stdio, with an MCP Bundle available for Claude Desktop. It follows robots.txt policies by default and offers an override for testing.