Booking.com MCP Server
以下摘录自该仓库的 README(片段) · 在 GitHub 读完整 README →
A hosted Model Context Protocol (MCP) server that gives Claude, Cursor, Windsurf and any other MCP client two read-only Booking.com tools. Search stays by destination and dates with rich filters, and read a single property in full, all as structured JSON, with nothing to host.
It reads public property pages on Booking.com that a signed-out visitor can see.
**1,000 free credits every month, no card required**, which is 100 Booking.
- Prompts and resources
Contents
- What you need
- Quick start
- Example prompts
- Tools
- Errors and failure paths
- Pricing, free tier and limits
- Tool selection
- How it compares
- FAQ
- HasData links
- Development
- Contributing
- License
What you need
An MCP client and a HasData API key from the dashboard, free to create with no card, and the free tier covers 100 calls a month at the 10-credit rate. This is a remote server, so the simplest path is a URL and an x-api-key header, with no container to run. A client that only speaks stdio reaches it through a thin launcher, published as @hasdata/booking-mcp on npm and hasdata-booking-mcp on PyPI, shown below.
Quick start
The server URL is the same for every client. We run it hands-on in Claude Code and Claude Desktop. The other blocks follow each client's own documented format for a remote server.
| Field | Value |
| :--- | :--- |
| URL | https://mcp.hasdata.com/mcp?apis=booking |
| Transport | HTTP, streamable |
| Auth header | x-api-key: HASDATA_API_KEY |
Clients with OAuth support can add the same URL as a connector and sign in without putting a key in a config file.
Claude Code
Claude Desktop
Settings, then Connectors, then Add custom connector, then paste https://mcp.hasdata.com/mcp?apis=booking and sign in.
For the config-file route, Claude Desktop loads only local (stdio) servers, so it reaches a remote server through a stdio launcher. The @hasdata/booking-mcp package is that launcher, and it reads the key from the environment. Add this to claude_desktop_config.json:
For Python instead of Node, swap the launcher for the PyPI package, which uvx runs without a manual install:
Cursor
~/.cursor/mcp.json for every project, or .cursor/mcp.json for one:
Windsurf
~/.codeium/windsurf/mcp_config.json. Windsurf calls the field serverUrl, not url:
VS Code
.vscode/mcp.json in the workspace:
Example prompts
Prompts, not code. Paste one in and the agent picks the tool itself. Each is annotated with the calls it takes, because every successful call costs 10 credits.
> Search Booking.com for hotels in Paris from September 15 to 18 for two adults, and give me the ten best-reviewed under $700 for the stay.
*One call, 10 credits. Price, review score and location come back on the search result.*
> Take the top result and pull its full detail: facilities, house rules, the room options, and the category ratings.
*One call, 10 credits. Those live on the property page, which the details tool reads by URL and dates.*
> Find four-star hotels in Paris with free cancellation near the center, and list price and review score.
*One call, 10 credits. Star rating, cancellation policy and distance are filters on the one request.*
> Compare the cheapest stay in Paris against Rome for the same dates.
*Two calls, 20 credits, one search per city.*
The property tool needs the same dates and guest counts as the search, because availability and price depend on the window. A search to shortlist plus a detail call on three properties is one search and three property calls.
Tools
| Tool | What it returns |
| --- | --- |
| hasdata_booking_place_getBookingPlaceDetails | The property identity (hotelId, title, address, coordinates), policies (free cancellation, no prepayment, child/pet stays), price, rating and review summary, photos, and…. 10 credits a call |
| hasdata_booking_search_getBookingSearchResults | Each hotel's hotelId, title and Booking URL, location info (city, address, coordinates, distance to center / nearest beach), policies (free cancellation, no…. 10 credits a call |
Two tools, read-only. Samples below are trimmed from real calls, and prices move constantly. Read them as shapes. Each tool name links to its endpoint reference, which carries the full field list.