Overview

The Triplyfy MCP Server implements a Model Context Protocol (MCP) backend tailored for travel itinerary management. It provides a compact set of REST/JSON tools that let LLM-based agents and client apps create and manipulate itineraries, place markers, rich notes, interactive maps and flight search/save records. By exposing these capabilities as programmatic endpoints, Triplyfy makes it straightforward for conversational agents to persist context, augment user plans, and coordinate multi-step travel workflows.

This server is useful for developers building travel assistants, booking flows, or map-driven experiences that rely on LLMs to generate, modify, or summarize travel plans. Instead of embedding state only inside an LLM prompt, Triplyfy provides durable structured state — itineraries, POIs, notes, and flight records — which can be queried, updated, and referenced during the conversation lifecycle.

Features

Create, read, update and delete travel itineraries
Add and manage places, attractions and custom POIs on itineraries
Attach searchable notes and media to itinerary items
Interactive maps support (geo coordinates, bounds, markers)
Flight search and save functionality to store candidate flights
REST API designed to be consumed by LLMs as MCP tools
Local development and containerized deployment options
Simple JSON schemas for easy integration with frontend clients

Installation / Configuration

Clone, install dependencies, and run locally:

git clone https://github.com/helpful-AIs/triplyfy-mcp.git
cd triplyfy-mcp
npm install
npm run dev   # or `npm start` for production build

Environment variables (example .env):

PORT=8080
DATABASE_URL=postgres://user:pass@localhost:5432/triplyfy
REDIS_URL=redis://localhost:6379
LLM_API_KEY=sk-xxxx          # optional, if server forwards to a model API
FLIGHTS_API_KEY=yourFlightsKey # optional, for integrated flight search

Running with Docker:

# docker-compose.yml (excerpt)
version: "3.8"
services:
  triplyfy:
    image: helpfulais/triplyfy-mcp:latest
    ports:
      - "8080:8080"
    environment:
      - DATABASE_URL=${DATABASE_URL}
      - REDIS_URL=${REDIS_URL}
      - LLM_API_KEY=${LLM_API_KEY}

Start:

docker compose up -d

Available Tools / Resources

Triplyfy exposes a set of API endpoints intended to be used as MCP tools by LLMs or client apps. Below is a canonical list (paths are representative — check the repository for exact routes and request/response shapes).

Method	Path	Purpose
POST	/itineraries	Create a new itinerary
GET	/itineraries/:id	Retrieve an itinerary
PATCH	/itineraries/:id	Modify itinerary metadata (dates, title)
DELETE	/itineraries/:id	Delete an itinerary
POST	/itineraries/:id/places	Add a place or POI (name, coords, notes)
GET	/itineraries/:id/places	List places for itinerary
POST	/itineraries/:id/notes	Add structured notes or recommendations
POST	/flights/search	Search flights (query params: origin, dest, dates)
POST	/flights/:id/save	Save a selected flight to an itinerary
GET	/maps/:id	Get map configuration (bounds, markers)

Other useful artifacts you’ll often find in the repo:

JSON schema files for itinerary, place, flight objects
Example Postman/HTTP collection for rapid testing
Sample front-end snippets demonstrating map rendering

Use Cases

Single-user planning assistant:
- A user chats with an LLM to “plan a 5-day trip to Kyoto”.
- The LLM calls Triplyfy to create an itinerary, add daily place markers, and attach recommended restaurants and transit notes.
- The client renders an interactive map based on the itinerary map object.
Collaborative trip editing:
- Multiple participants share an itinerary link.
- Each participant’s changes (new places, notes, flight candidates) are persisted in Triplyfy and visible to others in real-time or on refresh.
Conversational booking flow:
- The LLM queries /flights/search for candidate flights, summarizes options to the user, and on confirmation saves the selected flight to the itinerary (POST /flights/:id/save).
- Saved flight metadata can be used by downstream services for reminders or calendar invites.
Agent augmentation for travel agents:
- An LLM-powered assistant uses Triplyfy as long-term memory for client preferences (notes, saved places).
- When generating recommendations, the agent reads past itinerary data to avoid suggesting places the client already visited.

Getting Help & Contributing

Source code and issues: https://github.com/helpful-AIs/triplyfy-mcp
If you want to extend the server (new tools, extra adapters for flight providers or map tiles), follow the repository contributing guidelines and open a PR with tests and schema documentation.

This server is intended as a practical MCP implementation for travel use cases — it centralizes itinerary state so LLMs can operate reliably across multi-turn flows and integrate with UI map components and booking services.

Triplyfy MCP Server: LLM Travel Itinerary Manager