mcpField
T
Traveler.md

Traveler.md

Remote
traveler.mdMaps & Travel0Updated 3w ago

About Traveler.md

Traveler.md is a hosted travel-memory MCP server for travelers and their assistants. It stores a durable preference profile and trip plans, but does not search, price, or book travel.

Key features

  1. 01Store travel preferences
  2. 02Create and update trip plans
  3. 03Search and archive trips
  4. 04Protect concurrent edits with version hashes

Use cases

  1. 01Recall preferences before travel recommendations
  2. 02Track dates, itineraries, bookings, and notes
  3. 03Save a new trip idea for later

MCP support

Exposes profile and trip management tools to remote MCP clients over streamable HTTP, with OAuth 2.1 authorization.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://mcp.traveler.md/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

Details

From the project README

Traveler.md MCP Server

Traveler.md gives an AI assistant a traveler's own travel memory. It holds a durable profile of how the traveler likes to travel and one plan per trip. The traveler owns the data, authorizes access once, and can revoke it at any time.

This repository describes the hosted MCP (Model Context Protocol) server at https://mcp.traveler.md/mcp. The server source is not published here. The full documentation is at https://docs.traveler.md/mcp.

What the server does

The server stores two kinds of document and returns them on request.

  • traveler.md is the profile. It holds stable preferences: seat choice, hotel style, budget, dining style, travel companions. One per traveler.
  • trip.md is one trip. It holds destinations, dates, itinerary, bookings and notes. Many per traveler.

An assistant reads these before it plans or recommends, so the traveler does not repeat themselves. An assistant writes back what it learns, so the next conversation starts with the same knowledge.

Scope: travel memory only. The server has no inventory, no prices, no availability and no recommendation engine. Your assistant does the recommending. The profile changes what the recommendation is based on.

Tools

ToolPurpose
read_profileReturn the profile, optionally filtered to named sections
create_profileCreate the profile on first use
update_profileChange one or more profile sections
list_tripsPage or search trips by title, content, status or date
read_tripReturn one trip
create_tripCreate a trip with a title and a status
update_tripChange a trip's status, dates, sections or dated events
archive_tripFile a trip away

Every document is a set of named sections. Each section is a list of short sentences. Updates require the version_hash from the most recent read, so two assistants writing the same record cannot overwrite each other silently.

Configuration

The server uses streamable HTTP and OAuth 2.1. There are no API keys and no environment variables. On first use the client opens a browser window. The traveler signs in to Traveler.md and approves the connection.

VS Code and GitHub Copilot

One click:

Add to VS Code

Or add this to .vscode/mcp.json in a workspace, or to your user mcp.json:

{
  "servers": {
    "travelermd": {
      "type": "http",
      "url": "https://mcp.traveler.md/mcp"
    }
  }
}
Other clients
claude mcp add --transport http travelermd https://mcp.traveler.md/mcp   # Claude Code
codex mcp add travelermd --url https://mcp.traveler.md/mcp               # Codex

Any client that speaks remote MCP takes the URL directly:

{
  "mcpServers": {
    "travelermd": {
      "url": "https://mcp.traveler.md/mcp"
    }
  }
}

Per-client guides for Claude, Cursor, ChatGPT, Gemini and others are at https://docs.traveler.md/mcp.

Usage

Once connected, ask about travel in plain language. The assistant calls the tools.

You sayThe assistant calls
"Where should we stay in Lisbon?"read_profile, then recommends
"What is my next trip?"list_trips sorted by start date
"What is the plan for Kyoto?"list_trips, then read_trip
"I always want an aisle seat"update_profile
"We booked the ryokan"update_trip
"I would love to see Patagonia one day"create_trip with status Dreaming

Assistants often answer a travel question from the conversation alone and leave the profile unread. Add a standing instruction to your assistant's instruction file so it reads the profile first:

Before you recommend, plan, shortlist or book anything travel-related, call read_profile on the travelermd server. When a specific trip is in play, find it with list_trips and open it with read_trip.

Skills and agent plugin

An Agent Plugin that packages this server together with skills for onboarding, trip interviews, trip organization and post-trip reviews is at https://github.com/TravelAiSolution/traveler-md-agent-plugin. The skills teach an assistant the section names, the sentence caps and the read-before-write loop.

Data and privacy

The traveler owns every record. Access is granted per client through OAuth and revoked from the Traveler.md account page. The server rejects payment card numbers, passport numbers, known traveler numbers, social security numbers and door or access codes. The tools are also instructed not to record health, accessibility, dietary or biometric details. The privacy policy is at https://traveler.md/privacy.

Support

Open an issue in this repository for questions about the server or its tools. For account or security matters, write to [email protected].

License

The README and the files in this repository are licensed under the MIT License. The hosted server is a Traveler.md service. Traveler.md and TravelAI are trademarks of TravelAI; this license grants no rights to use them.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.