AirTreks MCP Server is a complex itinerary route planning server launched by AirTreks, designed to provide AI agents with route analysis capabilities for multi-stop, round-the-world, open-jaw, and ground-segment itineraries that standard flight search engines cannot handle.
Purpose When a user proposes an itinerary involving more than 3 intercontinental stopovers, this server can provide route analysis based on real AirTreks fare construction data, helping determine which itineraries are feasible and which will fail at the ticketing stage, and providing carrier recommendations.
Main capabilities
- Multi-stop and round-the-world route planning: Evaluate the feasibility of multi-segment itineraries, returning carrier suggestions and alternatives for each flight segment.
- Alliance fare bookability analysis: Based on more than 1,400 real fare construction attempts, provide bookability rates by number of segments (91% for 3-4 segments, 61% for 5-6 segments, 6% for 7 or more segments), pointing out cases where alliance round-the-world fares may fail.
- Dead segment identification: Includes 53 known dead segments (city pairs that appear bookable but fail under alliance fares), as well as "poison carriers" and hub fix suggestions, helping avoid ticketing failures.
- Price estimates: Provide historical price ranges for routes via fare_product_match, and price complete multi-stop itineraries across multiple tickets via trip_idea_create, which can show that the same itinerary may have a threefold price difference depending on how it is ticketed.
- Real fare queries: Return historical price ranges for any route in less than a second, as well as live fares for specific itineraries.
- Ground segment suggestions: For example, suggest switching to land transport between certain city pairs to save costs and add countries passed through.
How to use
- As an MCP connector: In a supported client (such as Claude.ai), go to Settings → Connectors → Add custom connector, and paste https://mcp.airtreks.com/mcp and you are done. Cursor users can add the same JSON configuration in .cursor/mcp.json (Streamable HTTP).
- Run locally in stdio mode: You can run the server locally and connect via stdio.
- REST API: All tools are also available as ordinary REST endpoints, with JSON input and output, suitable for agent frameworks that consume REST rather than MCP. For the complete OpenAPI 3.1 specification, see https://mcp.airtreks.com/openapi.json。
- Usage example: Ask the agent, "Plan a round-the-world trip: San Francisco, Tokyo, Bangkok, Singapore, Delhi, Istanbul, London, then return to San Francisco," and call plan_route to get an analysis result including carrier suggestions for each segment, bookability rates, ground segment alternatives, and price ranges.
Use cases Suitable for users who need to plan multi-stop intercontinental itineraries, round-the-world trips, open-jaw trips, or complex travel including ground segments, such as individuals, families, or large groups. It can also be used for AI travel assistants that need to provide users with reliable route analysis and price expectations.
Limitations
- Free to use, 100 requests per IP per day, no API key required; after registration, limits can be increased via an API key (X-API-Key header or ?key= parameter).
- trip_idea_create submits real customer contact information and should only be called when the customer is informed and agrees; repeated submissions with the same email and route within 24 hours will return the existing request instead of creating a duplicate.
- Price ranges are honest estimates rather than quotes; exact prices depend on cabin availability on the day of booking.
- Advisor handoff is handled by AirTreks advisors for actual pricing and ticketing.
This server has accumulated experience building complex multi-stop itineraries since 1987, and its data comes from real fare construction failure cases and actual booking records, information that other flight tools typically do not provide.
