Public Portal for the sdmx-mcp-ng repo
SDMX-MCP-NG Public Portal
Access Gateway for the SDMX-MCP-NG Project
🏆 FOR HACKATHON PARTICIPANTS 🏆
Welcome to the State Hackathon 2025!
🚀 Try the Chat Interface
We've built a conversational AI chat interface that lets you query 873 Luxembourg statistical dataflows using natural language!
Quick Start:
- See Chat Interface README
- Browse Query Examples
- Read Hackathon Proposal
What you can do:
- 💬 Ask questions in multiple languages: English, German (Deutsch), French (Français), Luxembourgish (Lëtzebuergesch), and more!
- 📊 Get real SDMX data in 3-8 seconds
- 🔍 Discover 873 Luxembourg dataflows
- 📈 Analyze employment, GDP, prices, population data
Example queries:
- 🇬🇧 "Get unemployment statistics for 2024"
- 🇩🇪 "Zeige mir Arbeitslosigkeitsdaten für 2024"
- 🇫🇷 "Donne-moi les statistiques de chômage pour 2024"
- 🇱🇺 "Gëtt mir d'Aarbechtsloosenzuelen fir 2024"
System Status: ✅ All services operational and ready!
🚀 Quick Start: Deploying the Platform
This guide provides the fastest way to get the SDMX-MCP-NG platform running.
Step 1: Get Access to the Private Repository
The main source code is in a private repository. You must be granted access before you can deploy/play with the platform.
- ➡️ Click here to fill out the Project Access Request form
- Wait for approval (typically 1-3 business days) and accept the GitHub invitation you receive via email.
Step 2: Clone This Public Portal
Clone this repository to your local machine to get the deployment script.
git clone https://github.com/sdmx-mcp-ng-orga/sdmx-mcp-ng-public-portal.git
cd sdmx-mcp-ng-public-portal
Step 3: Run the Setup Script
Once you have been granted access, run the setup command. This script will check your permissions and clone the private sdmx-mcp-ng repository for you.
# Make the script executable
chmod +x deploy-public-portal.sh
# Run the setup
./deploy-public-portal.sh setup
The private repository will be cloned into a new sdmx-mcp-ng directory.
Step 4: Configure API Keys
The platform requires API keys for the LLM provider (like Gemini, Claude, or OpenAI).
- Navigate to the private repository: cd sdmx-mcp-ng
- Create a .envfile:cp .env.example .env
- Edit the .envfile (nano .env) and add yourPLANNER_LLM_API_KEYand set a secureNEO4J_PASSWORD.
Step 5: Deploy the Platform
Now you can start all the services using the deployment script.
# From the sdmx-mcp-ng-public-portal directory
./deploy-public-portal.sh dev
This command starts the entire stack (database, applications, monitoring) in development mode.
Managing the Environment
Use the wrapper script to manage your deployment:
- Stop all services: ./deploy-public-portal.sh down
- Check service status: ./deploy-public-portal.sh status
- View logs: ./deploy-public-portal.sh logs
- Check health: ./deploy-public-portal.sh health
📦 What is SDMX-MCP-NG?
An advanced, AI-driven platform for interacting with SDMX (Statistical Data and Metadata eXchange) data.
Key Features
- 🗄️ Neo4j Graph Database - Efficient metadata storage and querying. in a future a GraphQL driven architecture could be considred
- 🤖 AI-Powered Queries - Natural language to Cypher using LLMs
- 🔌 MCP Protocol - Standard integration with AI assistants
- 🚀 Fast Demo Setup - Pre-populated backups/golden images for quick start
- 🌍 Multi-Source Support - NSIs, ECB, and other SDMX providers
Technology Stack
- Database: Neo4j 5.21 with APOC
- Backend: Python 3.10+ with FastMCP
- Protocol: Model Context Protocol (MCP)
- AI/LLM: Google Gemini, Anthropic, Claude, OpenAI,Mistral, Custom
- Deployment: Docker & Docker Compose
📚 Documentation
Full documentation is available in the private repository after access is granted.
Available after access:
- Installation & setup guides
- Architecture documentation
- API reference
- Configuration guides
- Demo tutorials
🤝 Access Levels
Reader (Default)
- ✅ Clone and pull repository
- ✅ View code and documentation
- ✅ Create and comment on issues
- ❌ Push code
Writer (By Request)
- ✅ All Reader permissions
- ✅ Push to branches
- ✅ Create pull requests
Most users receive Reader access
💬 Contact & Support
- Access Requests: Use the issue template
- Questions: GitHub Discussions
- Response Time: 1-3 business days
❓ FAQ
Q: Is this open source? A: Yes in the future, meanwhile for the moment the private repository is with controlled access in order to refine (sdmx-graph db schema mapping, LLM Agentic Architecture, golden image to accelerate the deployment process ...) .
Q: How long does approval take? A: Typically 1-2 business days.
Q: Can I redistribute the code? A: No, others must request their own access.
Q: Is there a cost? A: No, access is free for legitimate use cases (research, development, evaluation).
📜 License
The code in the private repository has its own license, which is available after access is granted. This public portal repository is under the MIT License.