# AI integrations: MCP, WebMCP and the Alice skill

> Detailed instructions for searching flights via AI: WebMCP, server-side MCP (Model Context Protocol) and the Yandex Alice skill. Connect your own AI agent.

AviaSeek can be used from AI agents: flight search is available via the MCP protocol (server), through WebMCP in the browser and by voice via the Yandex Alice skill.

## How to find flights without the UI

- To find flights, open https://aviaseek.com/en/search?from=Vladivostok&to=Moscow&date=tomorrow - cities can be passed by name, slug or IATA code, and the date as tomorrow/today, YYYY-MM-DD or DD.MM.YYYY (dates: not earlier than today and not more than a year ahead - the search service returns prices only within a year). The response is HTML with a list of options and booking links.
- If the date is unknown, leave it empty: https://aviaseek.com/en/search?from=Moscow&to=Vladivostok - a route page with a price calendar opens.
- A single JSON request with ready booking links: https://aviaseek.com/en/api/flights?from=MOW&to=LED&date=2026-10-10&limit=5 - the response has results[].booking_url (valid for about 15 minutes).
- Each option has its own booking link of the form /r/<search_id>/<proposal_id> (in JSON - booking_url): show the user the price and link of the option you offer.
- MCP (JSON-RPC 2.0, Streamable HTTP): POST https://aviaseek.com/en/mcp, tool searchFlights. Example: {"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"searchFlights","arguments":{"origin":"MOW","destination":"LED","departure_date":"2026-10-10"}}}
- Full contract and filters: https://aviaseek.com/en/llms-full.txt; the short site map and MCP card - https://aviaseek.com/en/llms.txt and https://aviaseek.com/en/.well-known/mcp.json.

## MCP server (Model Context Protocol)

The server MCP endpoint exposes the searchFlights tool: flight search by city IATA codes, filters (direct only, up to a price, departure time, airline, baggage), sorting and pagination. Each option contains a booking link, and the response for the model comes as a ready markdown list.

- [MCP server endpoint](https://aviaseek.com/en/mcp) - Streamable HTTP, POST
- [Server card](https://aviaseek.com/en/.well-known/mcp.json)
- [Tool and filter description](https://aviaseek.com/en/llms-full.txt)

## WebMCP in the browser

On the site pages the search tool is published for browser agents via WebMCP: an agent can run a search and show the same options with booking links that the user sees.

## Yandex Alice skill

The voice scenario lets you check prices and find flights: Alice asks for cities and a date, and AviaSeek replies with options and takes you to the search page.

## Documents for agents

- [llms.txt - short site map](https://aviaseek.com/en/llms.txt)
- [llms-full.txt - contract, parameters and examples](https://aviaseek.com/en/llms-full.txt)
- [Catalog cities](https://aviaseek.com/en/cities)
- [Help & tips](https://aviaseek.com/en/faq)

---

Source: https://aviaseek.com/en/ai-integrations
Sitemap for agents: https://aviaseek.com/en/llms.txt (full description - https://aviaseek.com/en/llms-full.txt)
MCP server: https://aviaseek.com/en/mcp