This workflow contains community nodes that are only compatible with the self-hosted version of n8n.
Execution Video: https://youtu.be/qkZ6UaO7aCE
Webhook
)4 people from Germany to Bangkok @14th August 2025
AI Agent
)Type: LangChain Agent
Model: Google Gemini 2.5 Flash via Vertex AI
Prompt logic:
Returns:
A markdown + bullet-format response with:
MCP Client List Tool
)listTools()
phaseMCP Execute Tool
)$fromAI('Tool_Parameters')
Google Vertex Chat Model
)Code Node
)Purpose: Parses AI output to extract:
Handles:
Output: Clean JSON:
{
"tripInformation": {...},
"listings": [...],
"totalListings": X,
...
}
HTTP Request
)API: Amadeus Flight Offers API
Purpose: Searches live flight offers using:
Auth: OAuth2
Code Node
)Purpose:
Output:
{
"tripInformation": {...},
"listings": [...],
"allFlightOffers": [...]
}
Set Node
)Purpose:
Assigns final response fields into clean keys:
traveldetails
listings
flights
This end-to-end workflow is a fully autonomous travel query-to-itinerary engine. From a plain text like “4 people from Vijayawada to Bangkok @14th August 2025,” it:
No manual parsing, no frontend — just AI + APIs + automation.
NOTE: I JUST USED A COMMUNITY NODE "n8n-nodes-mcp" + UNOFFICIAL AIRBNB MCP