Show HN: TrainYatri – Using MCP to Power LLM Driven Indian Railway Queries

4 months ago 37

TrainYatri is a powerful Model Context Protocol (MCP) server tailored for Indian Railways. Built using the indianrailapi and FastMCP framework, it seamlessly integrates real-time railway data leveraging LLMs(Large Language Models) enabling users to track trains, search stations, check PNRs, and more all through intelligent AI tools within a clean, extensible design.

🌐 What is the Model Context Protocol (MCP)?

The Model Context Protocol (MCP) is an open standard developed by Anthropic that standardizes how applications provide context to LLMs. Think of MCP as the "USB-C port" for AI applications, offering a universal interface for connecting AI models to various data sources and tools.

MCP enables developers to build secure, two-way connections between their data sources and AI-powered tools. It allows AI models to connect directly with external data sources, enabling them to read from and write to connected applications.

Live Train Tracking: Get context-aware real-time train positions.
PNR Status Checks: Receive natural language interpretations of PNR statuses.
Full Schedule Retrieval: Access structured station data for train schedules.
Seat Availability: Obtain insights with quota/class breakdowns.
Fare Enquiries: Leverage external APIs for fare information.
Autocomplete Station Search: Utilize fuzzy matching for station searches.
Live Station Status: Get updates on nearby train arrivals and departures.

Multi-tool Support: Automatically route user queries to appropriate tools.
Prompt Fallback & Error Recovery: Ensure natural responses even in ambiguous scenarios.
Clean Context Management: Maintain structured context for LLMs (stations, quotas, etc.).
Structured Tool I/O: Designed for AI agents & Inspector.
Flexible Deployment: Easily run with stdio , sse or streamable-http transport.

trainyatri/ ├── tools/ # Core logic per feature │ ├── fare_enquiry.py │ ├── live_status.py │ ├── pnr_status.py │ ├── seat_availability.py │ ├── station_search.py │ ├── station_status.py │ └── train_schedule.py ├── prompts.py # Prompt logic ├── resources.py # Shared resources (station codes, quotas, etc.) ├── context.py # Context-aware extensions ├── main.py # MCP server entry point ├── .env # Secrets and API keys (not committed) ├── requirements.txt # Python dependencies

git clone https://github.com/aryan/trainyatri.git cd trainyatri

pip install -r requirements.txt

Create a .env file:

RAIL_API_KEY=your_indianrailapi_key MODEL_API_KEY=your_openrouter_key

Or, run with Inspector enabled:

Tool Description

get_pnr_status	Get PNR status
get_fare	Retrieve train fare
check_seat_availability	Seat availability by quota/class/date
get_train_schedule	Train schedule lookup
get_live_status	Real-time train position
get_live_station_status	See trains at a given station
search_station	Autocomplete railway station names

🧑‍💻 How Users Interact with TrainYatri

🔍 Example 1: PNR Status Check

User asks: “Can you tell me the status of PNR 8701234567?” What happens: The MCP client triggers the get_pnr_status tool on the TrainYatri server → fetches live data from Indian Railways → and the LLM replies: “Your train is confirmed. Coach S3, seat 42. Boarding at Pune Junction on June 3rd at 18:40.”

🛤️ Example 2: Track a Live Train

User asks: “Where is train 12951 right now?” What happens: The MCP client invokes the get_live_status tool → receives real-time train coordinates → and the LLM responds: “Train 12951 is currently between Vadodara and Surat, running 10 minutes late.”

🪑 Example 3: Check Seat Availability

User asks: “Any 3AC seats from Mumbai to Jaipur on June 15?” What happens: The check_seat_availability tool runs → parses availability by date/quota/class → and the LLM replies: “Yes, there are 16 seats available in 3AC on train 12955 under the General Quota.”

🛠️ VS Code & Development Setup

This project includes a .vscode folder with recommended settings and launch configurations for a smooth developer experience.

Launch MCP Server: Use the pre-configured launch tasks in VS Code to start the server directly (look for "Run MCP Server" or similar in the Run & Debug panel).
Environment Variables: The .env file is automatically loaded if you use the VS Code Python extension.
Debugging: Step through requests and tool logic with breakpoints in VS Code.

🔌 Integration & Client Connectivity

LLM Clients: The MCP server exposes a 'stdio' transport compatible with clients like Claude Desktop, and GitHub Copilot (with custom endpoints).
Use the mcp.json file to configure any client to use the server.

🧠 How the MCP Server Works

FastMCP Framework: Handles routing, tool registration, and context management for LLM-based workflows.
Tool Registration: Each feature (PNR, fare, etc.) is a tool in /tools and auto-registered on startup.
Context & Resources: Shared data (station codes, quotas, etc.) is loaded at startup and injected into tool logic as needed.
Prompt Handling: Custom prompts and error handling ensure robust, natural responses even for ambiguous queries.
Extensibility: Add new tools by creating a Python file in /tools and updating main.py if needed.