Content
# Gen AI Travel Agent (Multi-MCP Servers Integration, RAG, Agno, OpenAI, Google SerpAPI, Multiple Agents)
A holistic AI-powered travel planning application that integrates multiple Model Context Protocol (MCP) servers to create detailed, real-time travel itineraries. Combines flight search, accommodation booking, location services, and comprehensive travel planning in one seamless interface.
### Demo Video
<video width="600" controls>
<source src="https://github.com/alivnavc/Gen-AI-Travel-Agent/raw/master/MCP_GenAI_Travel_Planner.mp4" type="video/mp4">
Your browser does not support the video tag.
</video>
## Features
- **Real-time Flight Search**: Live flight data with prices, airlines, and booking links
- **Airbnb Integration**: Real accommodation listings with current pricing and availability
- **Google Maps Services**: Location services, directions, and distance calculations
- **Comprehensive Itineraries**: Detailed day-by-day travel plans with timing and costs
- **Budget Planning**: Complete cost breakdown and budget optimization
- **Multi-MCP Architecture**: Seamless integration of custom and built-in MCP servers
## Prerequisites
- Python 3.8+
- OpenAI API key
- Google Maps API key
- SerpAPI key
- uv package manager (recommended)
## Quick Start
### 1. Install Dependencies
```bash
# Using uv (recommended)
uv sync
# Using pip
pip install -r requirements.txt
```
### 2. Set Up Environment Variables
Create a `.env` file:
```env
OPENAI_API_KEY=your_openai_api_key
GOOGLE_MAPS_API_KEY=your_google_maps_api_key
SERPAPI_KEY=your_serpapi_key
```
### 3. Start Flight Search MCP Server
First, set up the Flight Search MCP server from our GitHub repository:
```bash
# Clone the Flight Search MCP server
git clone https://github.com/alivnavc/Fligh-Search-MCP-Server.git
cd flight-search-mcp
# Install dependencies
uv sync
# Start the MCP server
uv run flight_search_mcp.py
```
The server will run on `http://localhost:8001/mcp`
### 4. Run Travel Agent Application
```bash
# In a new terminal, run the Travel Agent
uv run app.py
```
## MCP Integration Architecture
This application demonstrates advanced MCP integration by combining:
### Agno MCP Servers
- **Airbnb MCP**: Real accommodation listings and pricing
- **Google Maps MCP**: Location services and navigation
- **Google Search Tools**: Web search for current information
### Custom Flight Search MCP Server
- **Flight Search**: Real-time flight data via SerpAPI
- **Airport Information**: IATA codes and airport details
- **Price Trends**: Optimal booking time analysis
### Integration Configuration
```python
multi_mcp_tools = MultiMCPTools(
commands=[
"npx -y @openbnb/mcp-server-airbnb --ignore-robots-txt",
"npx @gongrzhe/server-travelplanner-mcp"
],
urls=["http://localhost:8001/mcp"], # Custom Flight Search MCP
urls_transports=["streamable-http"],
env={
"GOOGLE_MAPS_API_KEY": google_maps_key,
"SERPAPI_KEY": os.getenv("SERPAPI_KEY", ""),
},
timeout_seconds=60,
)
```
## Usage
### 1. Enter Trip Details
- **Departure Airport**: IATA code (e.g., BOM, JFK, LAX)
- **Destination Airport**: IATA code (e.g., DEL, SFO, LHR)
- **Dates**: Start and return dates
- **Duration**: Number of days
- **Budget**: Total trip budget in USD
### 2. Set Preferences
- Describe travel style and interests
- Select quick preferences (Adventure, Relaxation, etc.)
- Specify any special requirements
### 3. Generate Itinerary
- AI agent uses all MCP tools to gather real-time data
- Creates comprehensive day-by-day itinerary
- Includes flight options, accommodation, activities, and costs
### 4. Export and Share
- Download itinerary as ICS calendar file
- Share with travel companions
- Access detailed cost breakdown
## API Keys Required
| Service | Purpose | Get Key |
|---------|---------|---------|
| **OpenAI** | AI agent for itinerary generation | [OpenAI Platform](https://platform.openai.com/api-keys) |
| **Google Maps** | Location services and navigation | [Google Cloud Console](https://console.cloud.google.com/apis/credentials) |
| **SerpAPI** | Real-time flight search data | [SerpAPI](https://serpapi.com/) |
## Architecture Overview
```
┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐
│ Streamlit UI │ │ Agno Agent │ │ MCP Servers │
│ │ │ │ │ │
│ • Trip Input │◄──►│ • Orchestration │◄──►│ • Flight Search │
│ • Preferences │ │ • AI Planning │ │ • Airbnb │
│ • Results │ │ • Tool Usage │ │ • Google Maps │
└─────────────────┘ └──────────────────┘ └─────────────────┘
```
## Development
### Project Structure
```
Travel Agent/
├── app.py # Main Streamlit application
├── requirements.txt # Python dependencies
├── README.md # This file
└── LICENSE # MIT License
```
### Adding New MCP Tools
1. **Create Custom MCP Server**: Follow the pattern in our Flight Search MCP repository
2. **Update MultiMCPTools**: Add new server to the configuration
3. **Update Agent Instructions**: Include new tools in the AI agent's capabilities
4. **Test Integration**: Verify all tools work together seamlessly
### Customization
- **Modify Agent Instructions**: Update the AI agent's behavior and capabilities
- **Add New Data Sources**: Integrate additional MCP servers for more data
- **Custom UI Components**: Extend the Streamlit interface with new features
- **Export Formats**: Add support for different itinerary export formats
## Troubleshooting
### Common Issues
1. **MCP Server Not Running**
```bash
# Check if Flight Search MCP server is running
curl http://localhost:8001/mcp
```
2. **Missing API Keys**
- Verify all environment variables are set
- Check API key validity and credits
3. **Port Conflicts**
```bash
# Check port usage
netstat -ano | findstr :8001 # Windows
lsof -i :8001 # Linux/Mac
```
### Debug Mode
Enable detailed logging by setting the log level in `app.py`:
```python
logging.basicConfig(level=logging.DEBUG)
```
## Contributing
1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request
## License
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
## Related Projects
- [Flight Search MCP Server](https://github.com/alivnavc/Fligh-Search-MCP-Server.git) - Custom MCP server for flight search
- [Agno Framework](https://github.com/agno-ai/agno) - AI agent framework
## Support
For issues and questions:
1. Check the troubleshooting section
2. Review the error logs
3. Open an issue on GitHub
## Architecture Diagram
```mermaid
graph TB
subgraph "Frontend Layer"
UI[Streamlit Web UI]
UI --> |User Input| Agent
UI --> |Display Results| Results[Itinerary & Calendar Export]
end
subgraph "AI Agent Layer"
Agent[Agno AI Agent]
Agent --> |Orchestrates| MCPTools[MultiMCPTools]
end
subgraph "MCP Integration Layer"
MCPTools --> |Commands| BuiltInMCP[Built-in Agno MCP Servers]
MCPTools --> |HTTP| CustomMCP[Custom Flight Search MCP]
end
subgraph "Built-in MCP Servers"
BuiltInMCP --> Airbnb[Airbnb MCP Server]
BuiltInMCP --> GoogleMaps[Google Maps MCP Server]
BuiltInMCP --> GoogleSearch[Google Search Tools]
end
subgraph "Custom MCP Server"
CustomMCP --> FlightSearch[Flight Search Tools]
CustomMCP --> AirportInfo[Airport Information]
CustomMCP --> PriceTrends[Price Trends]
end
subgraph "External APIs"
Airbnb --> |Real-time Data| SerpAPI[SerpAPI]
FlightSearch --> |Flight Data| SerpAPI
GoogleMaps --> |Location Services| GoogleAPI[Google Maps API]
GoogleSearch --> |Web Search| GoogleAPI
end
subgraph "Data Processing"
Agent --> |Processes| FlightData[Flight Information]
Agent --> |Processes| AccommodationData[Accommodation Data]
Agent --> |Processes| LocationData[Location & Navigation Data]
Agent --> |Generates| Itinerary[Comprehensive Travel Itinerary]
end
subgraph "Output Layer"
Itinerary --> CalendarExport[ICS Calendar Export]
Itinerary --> CostBreakdown[Cost Breakdown]
Itinerary --> DetailedPlan[Day-by-Day Plan]
end
style UI fill:#e1f5fe
style Agent fill:#f3e5f5
style MCPTools fill:#fff3e0
style CustomMCP fill:#e8f5e8
style BuiltInMCP fill:#fff8e1
style SerpAPI fill:#ffebee
style GoogleAPI fill:#e3f2fd
style Itinerary fill:#f1f8e9
```
