Case Study: Using MCP to Create a Travel AI Assistant to Plan an Indonesian Volcano Trip (Installation Tutorial Included)

A while ago, I planned to take my family to Surabaya, Indonesia to see the volcano on May 1st. As a typical J person, the itinerary planning must be detailed.

However, we found that directly using large models on the Internet for brute force search planning does not provide timely and accurate information, especially for dynamic information such as air tickets and hotels.

So I wrote an MCP server myself - SearchAPI MCP. By connecting to the SearchAPI.io service, I can let the AI ​​assistant directly call Google Maps, flights, hotels and other real-time information to help me plan.

It has now been submitted to the MCP Planet official website, and everyone can go there to search for matching tutorials.

The core value of SearchAPI MCP is to connect AI  assistants with the powerful Google service ecosystem.

It allows AI to access multiple Google services in real time:

1. Google Search:  Get web results, knowledge graph information, search suggestions, etc.

2. Google Video:  Search videos, filter by duration, source, etc.

3. Google Maps:  Search for places and services, get addresses, reviews, coordinates and other detailed information.

4. Google Flights:  Check real-time flights, view price calendars, and filter flights as needed.

5. Google Hotels:  Search for hotels by location, check real-time prices, filter by facilities, and view ratings.

Through Search API, my AI assistant can help me search for real-time information on the Internet, plan routes, and check hotels and flights. All the information that I used to need to search on different apps is now concentrated in the AI ​​assistant, which is convenient and saves trouble.

Enter "SearchAPI" in the search box on the MCP Planet official website (MCPmarket.cn) to find the MCP.

• Preparation: You need to have basic Python and Git environment:

1. Clone the project

git clone https://github.com/RmMargt/searchAPI-mcp.git

2. Enter the project (it is recommended to create and activate a Python virtual environment to avoid dependency conflicts)

cd searchAPI-mcp

python -m venv venv

3. Activate the environment

Linux/Mac

source venv/bin/activate

Windows

.\venv\Scripts\activate

4. Install dependencies

pip install -r requirements.txt

Reminder: I am new to GitHub, please forgive me if the project structure or documentation is unclear and welcome to correct me.

•  Configure in AI client - Take Cherry Studio as an example

After installing the MCP server code, configure it in the AI ​​client you use (such as Cherry Studio, FastGPT, LobeChat, Cursor, and other MCP-supported applications).

Here we take Cherry Studio as an example:

1. Enter the "MCP Server Configuration" page and click "Add MCP Configuration"

• Name: Customizable

Search API

• Command: Please  replace /path/to/searchAPI-mcp/  with the actual project path you cloned
  

uv run –with mcp [cli] /path/to/searchAPI-mcp/mcp_server.py

• Environment variables
  

SEARCHAPI_API_KEY = Your SeachAPI.io Key, you need to apply for it yourself

After configuration, click the upper right button "searchapi" and the server will start successfully and run normally.

Create a new dialog in Cherry Studio, and when selecting the  MCP  server, check  the "searchapi" you just configured . Now your  AI  assistant has the ability to call various Google search services.

You can ask AI to search for hotels in a specific area, ask for details about facilities, check prices based on dates, and even check user reviews.

For example, I asked the AI ​​assistant: Help me search for hotels around Bromo volcano in Surabaya, Indonesia.

It will reconfirm the check-in time, number of people, age of children and other details.

I fed the AI ​​all the information and it returned real-time information for nine hotels, including their distance from the volcano.