Jinko
Jinko
README
Hotel Booking MCP Server
We are the first MCP server to give access to 2 millions of hotels with shopping and booking capabilities. You can integrate our MCP and start to sell within minutes in your agents. The prod version is available on request by requesting it on our website https://www.jinko.so/. Feel free to reach out!
MCP (Model Context Protocol) is an open protocol that standardizes how applications provide context to LLMs - think of it as a USB-C port for AI applications, connecting models to external data sources and tools.
Features
- Session Creation: Create a new booking session with location normalization
- Search Hotels: Search for available hotels based on location, dates, hotel amenities
- Get Hotel Details: Get all rooms and rate available for a specific Hotel ID
- Book Hotel: Book a hotel by creating a quote and returning a payment link
Session Management
The server maintains a session to store hotel search results, location information, and user preferences. This allows for efficient retrieval of hotel details without making additional API calls. The session includes:
- Conversation ID for tracking
- User language preference
- Currency and country code settings
- Confirmed place for hotel searches
- Alternative place suggestions
- Hotel search results cache
Development
You can install the package globally:
npm install -g jinko-mcp
Or run it directly with npx:
npx jinko-mcp
Running the Server
Using npm
npm run start
Using npx (after publishing)
npx jinko-mcp
The MCP server uses stdio transport, which means it can be used directly with MCP clients that support this transport type.
Installation
You can install our MCP locally for now. Remote support is coming. For developers that want access to the production environment and earn commission on sales, please contact us via our website https://www.jinko.so/.
{
"mcpServers": {
"jinko-travel": {
"command": "npx",
"args": [
"jinko-mcp-dev@latest"
]
}
}
}
Tools
1. create-session
Create a new booking session and normalize place for hotel search.
Parameters:
place: Location where user wants to search for hotels (e.g., 'New York', 'Paris', 'Tokyo')raw_request(optional): Summary of user's requirements in free textlanguage(optional): The language used by user, following ISO 639 (e.g., 'en', 'fr', 'zh')currency(optional): Currency code (e.g., 'EUR', 'USD')country_code(optional): Country code (e.g., 'fr', 'us')
This tool creates a new session with a unique conversation ID and normalizes the place input for hotel searches. It returns the selected place, alternative place suggestions, and available facilities for filtering hotels.
2. search-hotels
Search for available hotels based on location, dates, and other criteria.
Parameters:
place_id(optional): Optional place ID to override the default selected placecheck_in_date: Check-in date (YYYY-MM-DD), default: '2025-06-25'check_out_date: Check-out date (YYYY-MM-DD), default: '2025-06-26'adults: Number of adults, default: 2children: Number of children, default: 0facilities(optional): Facility IDs to filter hotels by, the IDs can be inferred with facilities resource
The tool uses the place selected during session creation by default, but you can override it by providing a specific place_id from the alternative places.
3. get-hotel-details
Get detailed information about a specific hotel by ID.
Parameters:
hotel_id: ID of the hotel to get details for
This tool retrieves detailed information about a hotel found in the search results, including rooms, amenities, and policies.
4. book-hotel
Book a hotel by creating a quote and returning a payment link.
Parameters:
hotel_id: ID of the hotel to bookrate_id: ID of the room to book
This tool creates a booking quote for the specified hotel and room, and returns a payment link for the user to complete the booking.
API Endpoints
The server uses the following API endpoints from the Jinko Travel BFF:
/api/v1/hotels/places/autocomplete: Get place suggestions based on user input/api/v1/hotels/availability: Search for available hotels/api/v1/hotels/{hotel_id}: Get hotel details/api/v1/booking/quote/schedule: Schedule a quote/api/v1/booking/quote/pull/{quote_id}: Pull quote status and details
Facilities Data
The server includes built-in facilities data to provide information about hotel amenities. This data is now directly embedded in the codebase as a constant in const.ts, making it more efficient and eliminating the need for external file loading.
The facilities data includes translations in multiple languages:
- English (en)
- Spanish (es)
- Italian (it)
- Hebrew (he)
- Arabic (ar)
- German (de)
Each facility includes:
facility_id: Unique identifier for the facilityfacility: Name of the facility in Englishsort: Sort order for displaytranslation: Array of translations in different languages
When creating a session, the available facilities are returned as part of the response, allowing LLMs to use the appropriate facility IDs when filtering hotel searches.
Publishing to npm
To publish this package to npm, follow these steps:
-
Make sure you have an npm account and are logged in:
npm login -
Update the version number in package.json if needed (current version is 0.0.5):
npm version patch # or minor or major -
Build the project:
npm run build -
Publish to npm:
npm publish
After publishing, users can install and run the package using npm or npx as described in the Installation section.
Configuration
The server uses the following default configuration (defined in config.ts):
- API Base URL:
https://api.dev.jinkotravel.com - Default Market:
fr - Default Currency:
EUR - Default Country Code:
fr
These defaults can be overridden when creating a session by providing the appropriate parameters.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。