Airbnb MCP server: let your AI agent search stays by place and dates
Wield is an MCP server that gives your AI agent one Airbnb tool. The agent can search stays by place, check-in and check-out dates, guest count and currency. It gets back names, total prices, ratings and listing links. The search runs in your own browser with your own session. Your passwords and cookies never leave your computer.
Airbnb Search Airbnb stays for a place and dates
Read-only, no loginRuns in your browser only
Parameters (6)
| Metric | Wield recipe | Browser Use + Claude Sonnet 5.5 |
|---|---|---|
| Success | 89% | 100% |
| Median time | 1.4 s | 70 s |
| Cost per run | $0 | $0.194 |
9 runs each, 7 November 2026. How we measure
What your agent gets
A Wield recipe is a small public file that describes the search request Airbnb's own site makes. Your agent runs it through wield-mcp, the local MCP server on your computer.
| Recipe | What it does | Login | Status (7 Nov 2026) |
|---|---|---|---|
airbnb.com/search-stays | Searches stays for a place and dates: name, type, total price, rating, badges, coordinates and link | No | ok, 5 of 5 checks |
The recipe takes a location, check-in and check-out dates, the number of adults, a currency (EUR, USD, GBP or CHF) and an optional maximum price per night. The how-to guide shows example prompts.
How a request flows
- You ask your agent for stays, for example in Lisbon from 4 to 7 December.
- The agent calls
find_recipe, thenrunwithairbnb.com/search-stays. wield-mcppasses the run through a local bridge to the Wield extension in your browser.- The extension sends the request from an airbnb.com tab, with your own session.
- The results return to your agent. Nothing passes through wield.run.
What leaves your computer
- One search request to
www.airbnb.comper run. - Wield never reads, copies or sends your cookies or passwords.
- Results go from the extension to
wield-mcp, then to your AI app. Results are marked as website content, so your agent is told not to follow instructions found in them. - Logs stay on your computer. They record the method, host, path, status and timing. They never record request bodies or query values.
Recipes are data, not code, and no JavaScript from the registry runs in the page. The registry signs every recipe with Ed25519, and wield-mcp refuses a recipe that is unsigned or has changed. Wield sends no Cookie or Host headers, and it redacts derived tokens from results and logs.
Airbnb's terms, in plain words
Airbnb's robots.txt disallows AI training crawlers (ClaudeBot, anthropic-ai). Airbnb's terms prohibit scraping. Wield runs this recipe only when you ask, one search at a time from your own browser, with one request per search. The scope is set: you start each search, one search per call, from your own browser, with no bulk collection and no purchases. Booking stays on airbnb.com. Check the site's terms before you automate anything beyond your own searches.
Limits and what is not working yet
- In our benchmark, Wield passed 8 of 9 runs. That miss stays in the data.
- Version 1 of the recipe left some prices empty in one browser session. Version 2 reads all three price formats.
- If Airbnb changes its search page, the recipe can stop working until it is updated.
- The recipe is read only. It does not book stays.
Every run stops at 12 steps, 40 requests, 5 MB per response or 60 seconds. Pacing is set per site, so a single search cannot run away.
FAQ
Is there an official Airbnb MCP server?
Not that we know of as of 7 November 2026. Wield's recipe reads the same search the Airbnb website uses.
Which AI apps work with it?
Any MCP client. npx -y wield-mcp init sets up Claude Desktop, Claude Code, Cursor and Codex. Claude Desktop also has a one-click bundle on wield.run/install.
Does it need an Airbnb login?
No. The search needs no login. It uses the session your browser already has, and your password never leaves your computer.
How fast is it compared with a browser agent?
As of 7 November 2026, in our preliminary benchmark, the recipe passed 8 of 9 runs with a median of 1,374 ms and no model cost. Browser Use driving Claude Sonnet 5.5 passed 4 of 4 runs, with a median of 70.2 s and $0.19 per run. The full method is on the benchmark page.