Navifare MCP Server

Version: 0.2.0

Compare flight prices across booking sites to find the best deal. Users provide flight details (conversationally or via screenshot), and Navifare searches Kayak, Google Flights, Booking.com, Kiwi, eDreams, Opodo, and more to find cheaper alternatives.

Overview

This MCP server wraps the Navifare REST API as MCP tools for flight price comparison. It enables AI assistants like Claude and ChatGPT to help users find better prices for flights they've already discovered. Supports round-trip flights, all cabin classes, and returns ranked results with direct booking links.

Connection Details

• Transport: HTTP (remote server) or STDIO (local)
• Production endpoint: https://mcp.navifare.com/mcp
• Documentation: https://www.navifare.com/mcp (auto-updates from the MCP endpoint)
• Learn more about MCP: Model Context Protocol

Claude Desktop Configuration (remote)

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "navifare": {
      "url": "https://mcp.navifare.com/mcp"
    }
  }
}

Tools

format_flight_pricecheck_request

Title: Format Flight Request

Parse and format flight details from natural language text or transcribed image content. Extracts flight information (airlines, flight numbers, dates, airports, prices) and structures it for price comparison. Returns formatted flight data ready for flight_pricecheck, or requests missing information if incomplete.

Annotations:

• readOnlyHint: true - Tool only formats/parses data, no external calls
• destructiveHint: false - Tool does not modify or delete data

Input Schema:

• user_request (required): Flight details in natural language text. Include all available information: flight numbers, airlines, departure/arrival airports and times, dates, prices, passenger counts, and travel class.

Example Request:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "format_flight_pricecheck_request",
    "arguments": {
      "user_request": "I found an Aeroitalia flight XZ2020 from FCO to MXP leaving tomorrow at 19.31 and returning with XZ2021 MXP to FCO next Friday at 10.11. 1 adult, economy, the best fare was 221 EUR"
    }
  }
}

flight_pricecheck

Title: Flight Price Check

Search multiple booking sources to find better prices for a specific flight the user has already found. Compares prices across different booking platforms to find cheaper alternatives for the exact same flight details.

Annotations:

• readOnlyHint: false - Tool performs searches and may trigger external API calls
• destructiveHint: false - Tool does not modify or delete data

Input Schema:

• trip (required): Flight trip details including segments, passengers, and travel class
  • legs (required): Array of flight legs (one for outbound, one for return in round trips)
  • travelClass (required): ECONOMY, PREMIUM_ECONOMY, BUSINESS, or FIRST
  • adults (required): Number of adult passengers (minimum: 1)
  • children (required): Number of child passengers (minimum: 0)
  • infantsInSeat (required): Number of infants requiring a seat (minimum: 0)
  • infantsOnLap (required): Number of infants on lap (minimum: 0)
• source (required): Source identifier for the query (e.g., "Claude", "MCP", "ChatGPT")
• price (required): Reference price found by the user (e.g., "84.00", "200.50")
• currency (required): Three-letter ISO currency code (e.g., "EUR", "USD", "GBP")
• location (required): Two-letter ISO country code for user location (e.g., "ES", "IT", "US"). If unsure, default to "ZZ"

Example Request:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "flight_pricecheck",
    "arguments": {
      "trip": {
        "legs": [
          {
            "segments": [
              {
                "airline": "XZ",
                "flightNumber": "2020",
                "departureAirport": "MXP",
                "arrivalAirport": "FCO",
                "departureDate": "2025-12-16",
                "departureTime": "07:10",
                "arrivalTime": "08:25",
                "plusDays": 0
              }
            ]
          }
        ],
        "travelClass": "ECONOMY",
        "adults": 1,
        "children": 0,
        "infantsInSeat": 0,
        "infantsOnLap": 0
      },
      "source": "MCP",
      "price": "84",
      "currency": "EUR",
      "location": "IT"
    }
  }
}

Example Response (truncated):

{
  "message": "Flight price search completed! Found 9 result(s):",
  "searchResult": {
    "request_id": "R1rPXQxY",
    "status": "COMPLETED",
    "totalResults": 9,
    "results": [
      {
        "rank": 1,
        "price": "174.67 EUR",
        "website": "Farera",
        "bookingUrl": "https://farera.com/booking?token=...",
        "fareType": "Special Fare"
      },
      {
        "rank": 2,
        "price": "176.00 EUR",
        "website": "Kiwi",
        "bookingUrl": "https://www.kiwi.com/deep?...",
        "fareType": "Special Fare"
      },
      {
        "rank": 3,
        "price": "186.00 EUR",
        "website": "Google Flights",
        "bookingUrl": "https://www.google.com/travel/flights/...",
        "fareType": "Standard Fare"
      }
    ]
  },
  "status": "COMPLETED",
  "userOriginalPrice": "200.00",
  "userOriginalCurrency": "EUR"
}

Typical Workflow

1. User provides flight details (text or screenshot)
2. Call format_flight_pricecheck_request to parse into structured format
3. Call flight_pricecheck with the returned flightData
4. Present ranked results with booking links to the user

Limitations:

• Only round-trip flights are supported. One-way and open-jaw routes are not supported.
• Results depend on real-time availability from booking sites (typically 5-9 sources respond within 30 seconds).

Revenue Share

Revenue share available for qualified partners!

Earn commission on flight bookings made through your integration. Revenue share available for qualified partners.

Compliance

This server complies with Anthropic's MCP Directory Policy. See docs/COMPLIANCE.md for details.

Privacy Policy and Terms of Service: https://navifare.com/terms
Contact: [email protected]
Privacy Inquiries: [email protected]

Environment Variables

The only environment variable you typically need to document for this MCP server is:

• GEMINI_API_KEY (required): Google Gemini API key for natural language parsing (used when your integration relies on Gemini for text/image understanding).

Examples

See docs/USAGE_EXAMPLES.md for complete usage examples with detailed workflows.

Documentation

Most documentation is in the docs/ directory (and a few files like this README.md and REPOSITORY_STRUCTURE.md live at the repo root):

Getting Started

• docs/QUICKSTART.md - 5-minute quick start guide
• docs/LOCAL_DEPLOYMENT.md - Comprehensive local deployment guide
• docs/RENDER_DEPLOYMENT_GUIDE.md - Production deployment guide

Reference

• docs/COMPLIANCE.md - Anthropic MCP Directory compliance details
• docs/USAGE_EXAMPLES.md - Complete usage examples
• docs/CLAUDE_SUBMISSION_CHECKLIST.md - Claude submission checklist
• REPOSITORY_STRUCTURE.md - Repository organization

Additional Guides

• docs/NGROK_SETUP.md - ngrok setup for local testing
• docs/IMAGE_INPUT_GUIDE.md - Image handling guide

Testing

Test files are located in the test/ directory. See test/README.md for details about available tests.

Running Tests

# Test with MCP Inspector
npx @modelcontextprotocol/inspector node dist/index.js

# Run specific test
node test/test-mcp.js

Scripts

Helper scripts are in the scripts/ directory:

• scripts/start-local.sh - Start local development server
• scripts/deploy-to-render.sh - Prepare for Render deployment
• scripts/start-servers.sh - Start multiple servers
• scripts/stop-servers.sh - Stop running servers

Repository Structure

navifare-mcp/
├── src/              # TypeScript source code
├── dist/             # Compiled JavaScript (generated)
├── docs/             # Documentation
├── test/             # Test files
├── scripts/          # Helper scripts
└── [config files]    # package.json, Dockerfile, etc.

See REPOSITORY_STRUCTURE.md for detailed structure.

Support

• General Inquiries: [email protected]
• Privacy Inquiries: [email protected]
• GitHub Issues: https://github.com/navifare/navifare-mcp

Reference

• MCP Documentation
• From REST API to MCP Server