travel-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| SERPAPI_API_KEY | No | SerpAPI key used in containers where no Keychain is available. When set, it takes precedence over the Keychain; an empty value is an error. | |
| TRAVEL_MCP_HOST | No | Allowed Host header for HTTP mode. Required together with TRAVEL_MCP_PATH for travel-mcp-http. | |
| TRAVEL_MCP_PATH | No | Secret URL path for HTTP mode. Must be at least 32 URL-safe characters. Required together with TRAVEL_MCP_HOST for travel-mcp-http. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
Search Google Flights through SerpAPI. Every tool call uses one search from a limited monthly SerpAPI quota, so ask before sweeping many dates or airports. Prices are in EUR unless currency is set. Pass tokens from results back exactly as returned.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| search_flightsA | Search one-way or round-trip flights on Google Flights. origin and destination are IATA codes; separate several with commas, e.g. "BRU,CRL". With return_date the result lists outbound flights and get_next_flights(departure_token) lists the return flights. outbound_times and return_times are hour ranges, "6,12" for departure or "6,12,10,18" for departure and arrival. max_stops 0 is nonstop only; bags counts carry-on bags. Airline filters take 2-letter IATA codes or STAR_ALLIANCE, SKYTEAM, ONEWORLD. deep_search is slower but closer to the browser. |
| search_multi_cityA | Search a multi-city trip of two or more legs. Each leg has origin and destination (IATA codes), date (YYYY-MM-DD) and optional times (hour ranges as in search_flights). The result lists the first leg; get_next_flights(departure_token) lists the next one. |
| get_next_flightsA | List return flights, or the next multi-city leg, for an offer's departure_token. |
| get_booking_optionsB | List where, and at what price, an offer's booking_token itinerary can be booked. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 4 tools
Each tool has a distinct role: single/round-trip flight search, multi-city search, continuation of an existing search via token, and booking option retrieval. The boundary between search_flights and search_multi_city is clear from the trip type, and the token-based tools are explicitly scoped.
All tools use consistent snake_case verb_noun naming: search_flights, search_multi_city, get_next_flights, get_booking_options. The pattern is predictable and easy for an agent to follow.
Four tools are well-scoped for a focused flight-search server. Each tool earns its place in the search and booking-option workflow, with no redundant or filler tools.
The surface covers core flight search workflows: one-way/round-trip, multi-city, continuation, and booking options. Minor gaps exist around ancillary needs like airport lookup or explicit offer details, but core workflows are covered without dead ends.