MCP Server

Connect AI agents to Nuitee Connect over MCP (Model Context Protocol). MCP lets Cursor, Claude, ChatGPT, Windsurf, and other clients discover tools, call them in a structured way, and run multi-step hotel workflows without custom glue code per editor.

Nuitee Connect exposes two MCP servers. Pick by how much access the agent needs.

Choose a server

	MCP for AI Agents	MCP for Developer Integrations
URL	`https://mcp.liteapi.travel/api/mcp`	`https://mcp.liteapi.travel/api/knowledge`
Auth	API key (`X-API-Key`) or OAuth	None
Docs, OpenAPI catalog, product guides	Yes	Yes
Live search / rates / prebook / book	Yes	No
Best for	Agents that query or book on your account	Learning the product, grounding answers, least privilege

MCP for AI Agents — docs plus live API tools (search, rates, book) with an API key or OAuth.
MCP for Developer Integrations — documentation, OpenAPI catalog, and product guides only. No live booking tools. No API key.

Both servers share the same seven discovery tools (list_endpoints, describe_endpoint, search_docs, fetch_docs_page, fetch_documentation, search_dashboard_kb, fetch_dashboard_feature). MCP for AI Agents adds ~120 live execute tools that hit your account (hotels, booking, flights, experiences, eSIM, loyalty, vouchers, analytics, mapping).

Server	Tools (today)	Preview
MCP for AI Agents	~127	Discovery + live API tools such as `post_hotels_rates`, `post_rates_prebook`, `post_rates_book`, `post_flights_rates`
MCP for Developer Integrations	7	Docs / OpenAPI / product guides only — no booking tools

Full tables live on each child page. Call tools/list on the MCP URL for the live catalog.

When to use MCP vs other Building with AI surfaces

Goal	Use
Agent should call search, rates, or book	MCP for AI Agents
Agent should read docs / OpenAPI without booking	MCP for Developer Integrations
Agent should write a booking app in your repo	Skills for coding agents
One-shot paste to scaffold a hotel demo	Prompt for vibe coding tools

Keep production keys on the server. Never put them in client-side apps or public chat prompts.

Next steps

Open MCP for AI Agents or MCP for Developer Integrations for the endpoint, client config snippets, and tool list.
Point your MCP client at the URL (Streamable HTTP).
For MCP for AI Agents, send a sandbox key for sample calls, or your own key for your account.