Keago — the travel MCP for AI

What is an MCP?

MCP (Model Context Protocol) is an open protocol that lets a language model use tools and data that live outside it. A server publishes the tools; the assistant — Claude, ChatGPT, Cursor — connects to that server and can call them.

There is nothing to learn as a user: you connect once, and from then on your assistant can do things it could not do before.

What is a travel MCP?

A travel MCP is a server specialised in travel: destinations, day-by-day itineraries, stage routes, distances, stops, maps and accommodation. Instead of the model improvising prose, it writes a structured trip into a real tool.

Keago is that server. The endpoint is https://keago.app/mcp and it works in any MCP client.

How do I connect Claude to a travel planner?

From Claude in the browser, not the desktop or mobile app — the app has no option to add connectors. Open Settings → Connectors, choose Add connector → Custom connector, and paste the server URL.

1. Open claude.ai in a browser.
2. Settings → Connectors.
3. Add connector → Custom connector.
4. Paste https://keago.app/mcp and authorise with your Keago account.
5. Go back to Keago and ask for the trip.

How do I plan a trip with Claude?

With the connector in place you just talk: “I want a 15-day trip through Japan, flying into Tokyo and out of Osaka.” Claude reads the Keago catalogue, spreads the sights across days and saves the trip in your account with a share link.

Then you edit it in Keago: move stops, change days, add the hotel you already booked, upload your GPX track.

How is this different from an AI trip planner?

An AI trip planner puts a chat inside its own website — the AI is a feature of the product. A travel MCP does the opposite: the chat is the one you already use, and the product sits underneath as a tool. You change neither assistant nor habit.

The practical difference: the itinerary does not stay in a conversation you will lose. It becomes an editable trip with a map and a link.

What can a travel MCP do?

This is what Keago exposes to your assistant today:

1. Search destinations and stage routes in the catalogue, with links.
2. Read a city guide: what to see, suggested days, best season.
3. Create a trip with the day-by-day itinerary already built.
4. Add days, stops, notes and stays to an existing trip.
5. Build spot lists (eat, coffee, nightlife, see) and push them into the itinerary.
6. List, edit or delete your trips — deletion only on your explicit confirmation.

Is it free, and what can the assistant see?

Keago is free. The connector uses OAuth 2.1: you authorise once and every call runs as you, under row-level security. The assistant never sees passwords, nor anyone else's trips.

It also never invents hotels, flights or prices. Keago records what you decided and links out to search when you want it.

Human travel knowledge, already structured

Keago is not only the plumbing. The catalogue is city guides with the sights in a sensible order, stage routes with distance and elevation, and spot lists rated by people who went. That is what your AI reads when you ask it for a trip.

1. City guides — what to see and in how many days.
2. Stage routes — long trails and road trips, with a simulator.
3. Spot lists — eat, coffee, nightlife, see.