Travel Planning Assistant Using LangGraph And FastMCP
Successfully developed a Travel Planner Assistant powered by LangGraph and MCP, integrating real-time flights, hotels, weather, places, and timezone tools. Features modular MCP servers, OpenAI agent orchestration, and a Streamlit UI for end-to-end trip planning.
Ask AI about Travel Planning Assistant Using LangGraph And FastMCP
Powered by Claude Β· Grounded in docs
I know everything about Travel Planning Assistant Using LangGraph And FastMCP. Ask me about installation, configuration, usage, or troubleshooting.
0/500
Reviews
Documentation
π§³ Travel Planning Assistant using LangGraph + MCP
Successfully developed an intelligent Travel Planner Assistant powered by LangGraph, FastMCP, and Modular MCP Servers, providing real-time trip planning with flights, hotels, weather, places, and timezone informationβall orchestrated through an LLM agent and delivered via a clean Streamlit UI.
π Overview
This project implements a fully modular, agent-driven travel planning system capable of:
- Retrieving live flight options using the Amadeus API
- Searching hotel availability & pricing
- Fetching real-time weather for any city
- Recommending nearby attractions and points of interest
- Detecting timezones and local times
- Coordinating results through a LangGraph-powered LLM workflow
The system uses MCP (Model Context Protocol) servers for each data sourceβallowing plug-and-play scalability, clean separation of services, and fault-tolerant tool execution.
π§© Key Components
1. LangGraph Workflow
- Core agent loop implemented with LangGraph
- Integrated routing via
ToolNodeandtools_condition - Persistent thread state using SQLite checkpoints
- Handles user queries, tool execution, and final answer synthesis
2. MCP Servers (FastMCP-based)
Each microservice runs as an independent MCP server:
flights_mcp.pyβ Live flights via Amadeus APIhotels_mcp.pyβ Hotel search & filteringweather_mcp.pyβ Current weather + forecastplaces_mcp.pyβ Attractions & POIsmath_mcp.pyβ Utility math operations- Fully typed tool definitions with FastMCP decorators
3. Streamlit Interface
- Chat-style conversation interface
- Threads persisted and reloadable
- Supports user messages, agent messages, and tool results
- Real-time server logs shown inside Streamlit
π Features
- End-to-end travel planning from a single prompt
- Real-time data retrieval (flights, hotels, weather, timezone)
- LLM-based reasoning with multiple MCP tools
- Robust agent orchestration using LangGraph
- Stateless UI, stateful backend with saved chat threads
- Easily extendable (add new MCP tools in minutes)
π οΈ Tech Stack
- LangGraph β LLM agent workflow engine
- FastMCP + MCP β Modular tool servers
- LangChain β Message and tool abstractions
- Streamlit β Frontend UI
- Amadeus SDK β Flight data retrieval
- Geopy & TimezoneFinder β Geo and timezone utilities
- Python 3.10+
π Project Structure
βββ app.py # Streamlit frontend
βββ travel_planner_chatbot.py # LangGraph agent & workflow
βββ flights_mcp.py # Live flight search MCP server
βββ hotels_mcp.py # Hotel search MCP server
βββ weather_mcp.py # Weather MCP server
βββ places_mcp.py # Places & attractions MCP server
βββ math_mcp.py # Utility MCP server
βββ requirements.txt # Dependencies
βββ README.md
π§ How It Works
1. User sends a travel-related query through Streamlit.
2. LangGraph agent receives the message and evaluates needed tools.
3. Tools are executed via MCP servers running in separate processes.
4. Agent collects tool results, reasons over them, and generates a plan.
5. Final structured response is sent back to the UI.
π¦ Installation & Setup
Clone repo:
git clone https://github.com/your-username/travel-planning-assistant-using-fastmcp
cd travel-planning-assistant-using-fastmcp
Install requirements:
pip install -r requirements.txt
Start MCP servers:
python flights_mcp.py
python hotels_mcp.py
python weather_mcp.py
python places_mcp.py
python math_mcp.py
Run Streamlit app
streamlit run app.py
π€ Contributions
Pull requests are welcome! You can add more MCP tools (car rentals, restaurants, currency converters, maps, etc.) to extend functionality.
π License
MIT License.
