Market analytics
Occupancy, ADR, RevPAR, revenue, lead time, stay length and supply for any of 30,000+ markets, with p25–p90 percentiles and up to 60 months of history.
Remote MCP server · Streamable HTTP
Airbnb MCP Server for Short-Term Rental Data
Airbnb has no official MCP server. AirROI's Airbnb MCP server gives Claude, Codex, Cursor and any other MCP client 29 tools for occupancy, ADR, revenue, comps and live search rank across 190+ countries.
MCP protocol 2025-11-25Registry com.airroi/mcp
Works with
- Claude
- Codex
- Cursor
- VS Code
- Antigravity
- n8n
Terminalbash
claude mcp add --transport http airroi \
https://mcp.airroi.com \
--header "X-API-KEY: YOUR_API_KEY" --scope user~/.codex/config.tomltoml
[mcp_servers.airroi]
url = "https://mcp.airroi.com"
env_http_headers = { "X-API-KEY" = "AIRROI_API_KEY" }~/.cursor/mcp.jsonjson
{
"mcpServers": {
"airroi": {
"url": "https://mcp.airroi.com",
"headers": { "X-API-KEY": "${env:AIRROI_API_KEY}" }
}
}
}.vscode/mcp.jsonjson
{
"servers": {
"airroi": {
"type": "http",
"url": "https://mcp.airroi.com",
"headers": { "X-API-KEY": "YOUR_API_KEY" }
}
}
}Run in your terminal, then type /mcp.
Claude guide- Airbnb listings
- 20M+
- Countries
- 190+
- MCP tools
- 29
- Starting price per call
- $0.01
What is an Airbnb MCP server?
An "Airbnb MCP server" is any Model Context Protocol server that gives an AI assistant Airbnb-related tools.
Airbnb has no official one (as of September 2026), so every option is third-party, and each does one of three jobs: search listings to book, run your own listings, or analyze a market.
Airbnb has no public API either. Its only API, the Partner API, is invitation-only, for vetted property-management and channel-manager software. On September 23, 2026, Brian Chesky committed to an Airbnb AI agent in 2027: an agent, not an MCP server other assistants can use.
MCP itself is an open standard, governed since December 2025 by the Linux Foundation's Agentic AI Foundation. AirROI is the market-data kind of server: a hosted short-term rental MCP server that answers from a maintained dataset of 20M+ Airbnb listings.
| Your job | Kind of server | Examples | What it can't do |
|---|---|---|---|
| Find a place to stay | Listing-search scraper | Community open-source scrapers (run locally), Apify, Bright Data, HasData | No occupancy, revenue or history; breaks when Airbnb changes its pages |
| Run my own listings | Property-management or pricing server | Hospitable, Guesty (beta), Lodgify, dynamic-pricing tools; community Hostaway and OwnerRez servers | No data on markets or listings you don't manage |
| Analyze a market | Market-data server | AirROI | Cannot book stays, message guests or change prices |
- Planning a trip? Use a listing-search scraper, or search Airbnb itself.
- Running listings? Use your PMS's MCP server (Hospitable, Guesty, Lodgify) or your dynamic-pricing tool's.
- Deciding where to buy, benchmarking against comps, or reporting to owners? Use AirROI.
- Many hosts run two: a PMS server for their own bookings and AirROI for the market around them.
A real session: Austin vs Nashville
We asked AirROI's Airbnb MCP server one plain-English question: compare Austin and Nashville for a 2-bedroom entire home. The assistant made 6 tool calls costing $0.42, and every number in its answer traces to one of them.
The full session is below.
Nashville 2-bedroom entire homes out-earned Austin's by 61% over the trailing 12 months ($40,075 vs $24,965). Occupancy was nearly identical, so the whole gap came from rate. Each figure traces to an airroi_market_summary response you can open in the session.
The same question works for any pair of markets in 190+ countries. For hosts, property managers and revenue managers, see Airbnb MCP use cases.
Investor
Compare two markets before you buy
Prompt
Compare Austin and Nashville for a 2-bedroom entire-home Airbnb: occupancy, ADR, RevPAR, annual revenue, and how many competing listings I'd face.
6 tool calls
airroi_search_markets$0.01query: Austin
Response excerpt (first of 10 matches shown)
{ "entries": [ { "full_name": "Austin, Texas, United States", "country": "United States", "region": "Texas", "locality": "Austin", "district": "", "native_currency": "USD", "active_listings_count": 11888 } ] }airroi_search_markets$0.01query: Nashville
Response excerpt (first of 10 matches shown)
{ "entries": [ { "full_name": "Nashville-Davidson, Tennessee, United States", "country": "United States", "region": "Tennessee", "locality": "Nashville-Davidson", "district": "", "native_currency": "USD", "active_listings_count": 6639 } ] }airroi_market_summary$0.10Austin, Texas, United States · room_type = entire_home · bedrooms = 2 · num_months: 12 · currency: usd
Response excerpt
{ "market": { "country": "United States", "region": "Texas", "locality": "Austin", "district": null }, "occupancy": 0.48, "average_daily_rate": 235.7, "rev_par": 113.9, "revenue": 24964.8, "booking_lead_time": 36.7, "length_of_stay": 6, "min_nights": 7.7, "active_listings_count": 2028.8 }airroi_market_summary$0.10Nashville-Davidson, Tennessee, United States · room_type = entire_home · bedrooms = 2 · num_months: 12 · currency: usd
Response excerpt
{ "market": { "country": "United States", "region": "Tennessee", "locality": "Nashville-Davidson", "district": null }, "occupancy": 0.49, "average_daily_rate": 296.2, "rev_par": 146.3, "revenue": 40075.3, "booking_lead_time": 52.4, "length_of_stay": 3.8, "min_nights": 5.7, "active_listings_count": 1269.8 }airroi_market_revenue$0.10Austin, Texas, United States · room_type = entire_home · bedrooms = 2 · num_months: 12 · currency: usd
Response excerpt (2 of 12 monthly rows shown)
{ "market": { "country": "United States", "region": "Texas", "locality": "Austin", "district": null }, "results": [ { "date": "2025-10-01", "avg": 4524.8, "p25": 2237.8, "p50": 3940.5, "p75": 6055.7, "p90": 8689.1 }, { "date": "2026-01-01", "avg": 2656.4, "p25": 1138, "p50": 2354.2, "p75": 3773.1, "p90": 5248.1 } ] }airroi_market_revenue$0.10Nashville-Davidson, Tennessee, United States · room_type = entire_home · bedrooms = 2 · num_months: 12 · currency: usd
Response excerpt (2 of 12 monthly rows shown)
{ "market": { "country": "United States", "region": "Tennessee", "locality": "Nashville-Davidson", "district": null }, "results": [ { "date": "2025-10-01", "avg": 5731.5, "p25": 3269.6, "p50": 5009.2, "p75": 7344.1, "p90": 10455.2 }, { "date": "2026-01-01", "avg": 2637.3, "p25": 1148.3, "p50": 2171.6, "p75": 3618.5, "p90": 5229.8 } ] }
Answer written from these results
Over the trailing 12 months, 2-bedroom entire homes in Nashville earned more per listing: $40,075 in average annual revenue versus $24,965 in Austin. Occupancy is nearly identical (49% vs 48%), so the gap comes from rate: Nashville's ADR is $296 against Austin's $236, for RevPAR of $146 versus $114. Nashville also has less 2-bedroom supply, about 1,270 active listings a month versus about 2,029 in Austin. October 2025 was the strongest month for the median listing in both markets, at $5,009 in Nashville and $3,941 in Austin.
What you can ask
You can ask about any Airbnb market or listing. Occupancy is the share of all nights that were booked, ADR (average daily rate) is the average price per booked night, and RevPAR combines the two as revenue per available night.
The 29 tools also cover revenue, comps, booking pace, search rank and pricing.
Comps and revenue estimates
Projected annual revenue for any address, with a p25–p75 range, plus up to 25 comparable listings within 1–10 miles and what each one earned.
Live Airbnb data
Five live tools read a listing's calendar, rates, availability and cleaning fee when you ask, and rank up to 270 listings in Airbnb search order.
Price recommendations
A base price with conservative, balanced and aggressive options, and nightly rates that follow your pricing and stay rules. Nothing is ever published.
Bulk search and export
Search by market, a radius of up to 100 miles or a polygon you draw, then export up to 100,000 listings as CSV or JSONL.
Works in every client
One remote URL and one X-API-KEY header work in Claude, Codex, Cursor, VS Code, n8n and your own agents. There is nothing to install.
Questions a listing scraper can't answer
A scraper sees what a guest sees: asking prices and open dates. Here is how each kind of server handles six common questions.
| Question | Listing-search scraper | AirROI tool and price per call |
|---|---|---|
| Find me a 2BR in Lisbon for Oct 12–15 under €150 | YesIts purpose | NoNot a booking tool |
| Trailing-12-month occupancy for 2BR homes in Nashville | NoSearch pages show asking prices, not bookings |
|
| What would a 2BR/2BA at this address earn? | NoNo revenue data |
|
| My 25 nearest comps and what they earned | PartlyNearby listings, no revenue |
|
| Monthly trend over the last 5 years | NoA scrape is a one-time snapshot |
|
| Where do listings rank in Airbnb search for this map area? | PartlyOne results page for a text query |
|
How the AirROI MCP server answers a question
You ask in plain English, your assistant picks the tools, and the AirROI MCP server returns numbers from its dataset. Every request goes to one remote endpoint, mcp.airroi.com, and carries your API key in an X-API-KEY header.
Nothing runs on your machine, and the answer names the market, filter and period it used.
You ask in plain English
“Compare Austin and Nashville for a 2-bedroom.” No tool names or query syntax.
The assistant picks tools
Your client reads the 29 tool schemas from
tools/list, which isn't billed, and the model chooses the calls.It calls mcp.airroi.com
Each call carries your API key in the
X-API-KEYheader and returns JSON with averages and p25–p90 percentiles.You get a checkable answer
The reply names the market, filter and months covered, so every number traces to a tool result.
- Endpoint
- mcp.airroi.com
- Transport
- Streamable HTTP
- Sessions
- Stateless
- Auth
- X-API-KEY header
- MCP protocol
- 2025-11-25
- Registry
- com.airroi/mcp
Live scraping vs a maintained dataset
Scraper-based Airbnb MCP servers load airbnb.com pages each time you ask, so whenever Airbnb changes how those pages load, the tools can stop working. That suits “find me a place in Lisbon next week.”
It is harder to rely on for numbers you need to trust or repeat, for the reasons below.
Community scraper servers are free and run on your machine; hosted scrapers bill per request. Either way, a search page lacks what investors need: it shows asking prices and open dates, not how often a listing books or what it earned, and a single scrape has no history.
AirROI answers from stored market metrics, so your AI client never requests airbnb.com, and 5 live tools cover “right now” questions on request. Every failure mode and price is side by side in Compare Airbnb MCP servers.
| What can break | Why a dataset-backed server avoids it |
|---|---|
| Search comes back empty after Airbnb changes its search pages | Answers come from stored data, not a live search page |
| Fields such as amenities come back blank when Airbnb moves them into client-side code | Page layout changes never reach the AI client |
| Requests from non-US IP addresses can get a redirect page instead of results | The same answer from any country |
What the server can and can't touch
The AirROI MCP server answers questions; it never acts on Airbnb. It has no Airbnb login, never connects to your Airbnb account, and none of its 29 tools can change a listing, price, booking or message.
Your assistant never loads airbnb.com from your IP address either. The AI drafts; you decide what to do with the numbers.
What it can do
- Read market metrics, listing performance and comps anywhere
- Read live calendars, rates, availability and search rank on request
- Recommend a base price and nightly rates for you to review
- Create an export file or a scan cursor for you to use
What it can't do
- Sign in to Airbnb or to your host account
- Change a listing, a price, a booking or a message
- Publish rates, book stays or contact guests
- Load airbnb.com from your computer or IP address
26 of the 29 tools are flagged read-only. The other three (airroi_batch_listings, airroi_live_scan, airroi_export) create a download file or a scan cursor and write nothing to Airbnb. To act on your own listings, run your PMS's MCP server alongside AirROI.
What it costs
Tool calls start at $0.01, the same prices as the matching REST endpoints, paid from prepaid credits after a $10 minimum deposit. There is no subscription.
Listing the tools is free, and every request needs your API key. The Austin vs Nashville comparison above cost $0.42.
Market lookups cost $0.01; single market metrics and comps $0.10; pacing, live calendars and revenue estimates $0.20; and listing searches $0.50. Preferred-tier accounts pay about half. The eight real sessions we published cost $1.85 in total. See per-call pricing for every endpoint.
Two tools recommend a base price and a full nightly calendar, and neither publishes anything. Each costs $0.10 a call and explains its reasoning. See the price recommendation tools.
Compare two markets (Austin vs Nashville)
airroi_search_markets × 2$0.02airroi_market_summary × 2$0.20airroi_market_revenue × 2$0.20
Total$0.42
Typical 6-call market analysis
airroi_search_markets$0.01airroi_market_summary$0.10airroi_market_occupancy$0.10airroi_market_adr$0.10airroi_market_revenue$0.10airroi_market_future_pacing$0.20
Total$0.61
Standard per-call prices; listing the tools (tools/list) is not billed. Pay-as-you-go credits with a $10 minimum deposit.
Airbnb MCP vs Airbnb API vs AirROI API
Airbnb's own API is for approved software partners managing their listings. An “Airbnb MCP server” is always third-party.
AirROI's API and MCP server return the same market data: the API for your code, the MCP server for your AI assistant, at the same per-call price and with the same API key.
Scraper servers don't use the Airbnb API; they load public pages. PMS servers use it indirectly, through the PMS's own Partner API connection. AirROI's market tools read AirROI's own dataset, which your code can also reach through the AirROI REST API.
| Compared on | "Airbnb MCP server" (third-party) | Airbnb API (Partner API) | AirROI REST API | AirROI MCP server |
|---|---|---|---|---|
| Made by | Community or vendors | Airbnb | AirROI | AirROI |
| Who can use it | Anyone | Invitation only: vetted PMS, channel-manager and experience partners | Anyone with a key | Anyone with a key |
| Data | Public search pages (scrapers) or your own bookings (PMS servers) | Your own listings, calendars, reservations, pricing, messaging | Listing and market analytics for any market | Same data as the REST API |
| Reads or writes | Scrapers read; PMS servers read and write your account | Read and write your own listings | Read | Read; no tool changes a listing, price, booking or message |
| Built for | AI assistants | Software integrations | Your code, spreadsheets, Zapier and Make | AI assistants and agents |
| Interface | MCP tools | REST | REST (JSON) | 29 MCP tools over Streamable HTTP |
| Auth | None (local), OAuth or vendor token | Partner credentials after review | X-API-KEY | X-API-KEY |
| Price | Free to subscription | Not sold; partnership | From $0.01 per call | Same per-call prices as REST |
| Pick it when | You want to search stays, or act on your PMS account | You are a PMS or channel manager | You're building software or reports | You want answers in Claude, Codex, Cursor or an agent |
Works in the AI tools you already use
One URL and one header connect the AirROI MCP server to any client that supports remote HTTP servers with custom headers: Claude Code, Claude Desktop, Codex, Cursor, VS Code, Devin Desktop, Antigravity CLI, n8n and the OpenAI Agents SDK.
ChatGPT on the web is the exception; its developer mode accepts only OAuth or no-auth servers.
- Claude
- Codex
- Cursor
- VS Code
- Antigravity CLI
Codex works in the ChatGPT desktop app, CLI and IDE extension, which all send headers from ~/.codex/config.toml. Step-by-step guides: Connect Claude to Airbnb data; Use Airbnb data in Codex and ChatGPT; and the Airbnb MCP setup guide for Cursor, VS Code, n8n and every other client.
| Client | Status | How to connect | Watch out for |
|---|---|---|---|
Claude Code Connect Claude to Airbnb data | Works | claude mcp add --transport http … --header | Needs a Pro, Max, Team, Enterprise or Console account |
Claude Desktop Connect Claude to Airbnb data | Works via mcp-remote | mcp-remote bridge in claude_desktop_config.json | No spaces around the colon in --header; needs Node.js; restart the app |
Codex in the ChatGPT desktop app Use Airbnb data in Codex and ChatGPT | Works | Settings → MCP servers → Add server, or edit the TOML | Same file as the CLI; restart after editing |
Cursor Airbnb MCP setup guide | Works | mcp.json with url + headers | Enable the server and check its tools under Customize in the sidebar |
VS Code (Copilot agent mode) Airbnb MCP setup guide | Works | .vscode/mcp.json with "servers" and "type": "http" | The key is servers, not mcpServers |
n8n Airbnb MCP setup guide | Works | MCP Client Tool node → HTTP Streamable → Header Auth | Limit the agent to the tools the workflow needs |
ChatGPT on the web (developer mode) Use Airbnb data in Codex and ChatGPT | Not supported | Not supported for API-key servers | Doesn't read config.toml; use Codex instead |
Frequently asked questions
An MCP server that gives an AI assistant Airbnb-related tools over the Model Context Protocol, the open standard now governed by the Linux Foundation's Agentic AI Foundation. They come in three kinds: scrapers that search listings for travelers, PMS servers that act on a host's own account, and market-data servers. AirROI's is the market-data kind: 29 tools at https://mcp.airroi.com covering 190+ countries.
Anything about Airbnb market or listing performance: occupancy, ADR, RevPAR, revenue, comps, revenue estimates, booking pace, search rank or recommended prices. Asked to "Compare Austin and Nashville for a 2-bedroom," the server returned $40,075 average annual revenue in Nashville against $24,965 in Austin, at 49% and 48% occupancy. The use-cases page walks through investor, host and property-manager sessions.
You pay per tool call, from $0.01, the same as the REST API, after a $10 minimum deposit. There's no subscription. A typical 6-call market analysis (market lookup, summary, occupancy, ADR, revenue and forward pacing) costs about $0.61, and the eight real sessions we published cost $1.85 in total. Listing the tools (tools/list) is free.
No. AirROI's server is hosted at https://mcp.airroi.com and speaks Streamable HTTP, so there is nothing to run locally. Most clients take the URL plus an X-API-KEY header. Claude Desktop needs the small mcp-remote bridge through npx. Community scraper servers, by contrast, often run on your machine and load airbnb.com from your IP.
No. None of the 29 tools can change a listing, a price, a booking or a message, and AirROI never connects to your Airbnb account. 26 tools are flagged read-only. The other three only create a download file or a scan cursor. To act on your own listings, add your PMS's or pricing tool's MCP server alongside AirROI.
Any MCP client that can call a remote HTTP server with a custom header. That includes Claude Code, Claude Desktop, Codex (in the ChatGPT desktop app, the CLI and the IDE), Cursor, VS Code, Devin Desktop (formerly Windsurf), Antigravity CLI, n8n and the OpenAI Agents SDK. ChatGPT on the web is the exception, because its developer mode accepts only OAuth or no-auth servers. The setup page has the config for each.
Both names describe the same thing. "Short-term rental MCP server" and "vacation rental MCP server" are the category; the data behind AirROI's server is Airbnb listing and market data across 30,000+ markets. Hosts, property managers and investors all use it the same way: ask about a market, get the numbers.
Market and listing metrics run through the latest monthly snapshot, updated each month. The five live tools read Airbnb calendars, rates, availability and search rank at the moment you ask. Every answer states its market and period, so you can see what it's based on.
Yes. Each MCP tool maps to a REST endpoint and costs the same per call, so one key and one balance cover both. Use the REST API when you're writing code or feeding a spreadsheet. Use the MCP server when you want Claude, Codex, Cursor or another agent to decide which calls to make.
Airbnb's API, the Partner API, is invitation-only. It lets approved property-management and channel-manager software read and write their own listings, calendars, reservations and messages. An "Airbnb MCP server" is always third-party and gives an AI assistant tools instead. AirROI offers both over the same market data: the REST API for your code and the MCP server for Claude, Codex, Cursor or any other agent, at the same per-call price.
Occupancy is the share of all nights that were booked, with blocked nights counted as unbooked. ADR (average daily rate) is the average price per booked night. RevPAR (revenue per available rental night) combines the two, roughly ADR × occupancy. In a real session for 2-bedroom entire homes in Nashville, the server returned 49% occupancy, a $296 ADR and $146 RevPAR. Every market tool also returns p25, p50, p75 and p90 values.
Yes. It's listed as com.airroi/mcp in the official MCP Registry run by the Model Context Protocol project, pointing to the remote endpoint https://mcp.airroi.com over Streamable HTTP. Clients and directories that read the registry can find it there. You still add your X-API-KEY header when you connect.
Ask your first market question today
Get your API key, add one URL and one header to Claude, Codex, Cursor or any MCP client, and ask about any market in 190+ countries.