SkyGrouper is a MCP-based AI-Agent service that suggests optimal travel destinations and flight itineraries for a group of travelers departing from different airports. It combines user interests with budget constraints and leverages an Agent for detailed planning.

• SkyScanner MCP Server
• Shortlisting of candidate cities based on group interests
• Concurrent LLM-powered destination overview and flight plan generation
• Multi-origin support with individual budgets
• MongoDB integration for storing and retrieving trip requests

• Python 3.8+
• MongoDB (local or hosted)
• OpenAI API credentials (configured via agent_helpers)

1. Clone the repository:

   git clone https://github.com/jc2409/SkyGrouper.git
   cd SkyGrouper

2. Create a virtual environment and install dependencies:

   uv init
   source .venv/bin/activate

1. Copy the sample environment file and set your variables:

   cp .env.example .env

2. In .env, configure:

    SKYSCANNER_API_KEY=""
    OPENAI_API_KEY=""
    MONGO_URI=""

Insert a sample trip document for testing:

db.groupTrips.insertOne({
  createdAt: new Date(),
  users: [
    { from: "LHR", dates: { start: "2025-06-01", end: "2025-06-10" }, budget: { max: 1500 }, interests: ["culture","mountain"] },
    { from: "KOR", dates: { start: "2025-06-01", end: "2025-06-10" }, budget: { max: 1500 }, interests: ["beach","party"] }
  ]
});

cd agent/mcp
uv run server.py

cd ~
cd agent && uv run app.py

Fetches the latest trip request from MongoDB, generates a shortlist, and returns detailed plans.

curl -X POST http://127.0.0.1:7000/plan-trip \
     -H "Content-Type: application/json"

{
  "plans": [ /* array of { destination, flights, totals } */ ],
  "shortlist": [ /* array of candidate cities */ ]
}

• For manual testing, use the provided curl example after seeding Mongo.

This project is licensed under the MIT License. See LICENSE for details.

Developed during the HackUPC Hackthon in Barcelona