DataHub Data Literacy Game

• 

  Local-DataHub-Instance

• 

  system-architecture-diagram

• 

  Game-UI

Inspiration

Metadata literacy is essential for anyone working with modern data systems, yet most people learn it through documentation, dashboards, or troubleshooting. I wanted to create something more engaging — a way to experience metadata concepts instead of just reading about them.

The theme of Build with DataHub: The Agent Hackathon inspired me to merge two worlds: agentic systems and gamified learning. This led to the idea of a tunnel‑exploration game where an AI agent teaches metadata concepts through real interactions connected to DataHub.

What it does

datahub-data-literacy-game is an interactive learning experience that turns metadata literacy into gameplay.
Players explore a tunnel world, trigger events, and interact with a DataHub MCP agent that:

• Explains metadata concepts
  
• Shows ownership, lineage, and dataset context
  
• Performs real metadata updates in DataHub
  
• Connects gameplay actions to real DataHub operations
  

It transforms abstract metadata concepts into visual, hands‑on learning.

How we built it

The project combines three main components:

Game World

A lightweight tunnel‑exploration environment built with Python and Pygame. Players move through the world, triggering metadata learning events.

Agent Layer

A DataHub MCP agent that responds to gameplay actions, provides explanations, and performs real metadata updates. It acts as the bridge between the game and DataHub.

DataHub Integration

Gameplay events map directly to DataHub operations such as updating ownership, viewing lineage, inspecting metadata fields, and triggering context‑aware agent responses.

Challenges we ran into

Building an agentic game connected to a metadata platform came with several challenges:

• Designing gameplay triggers that map cleanly to metadata concepts
  
• Ensuring the MCP agent responds quickly without blocking the game loop
  
• Creating educational explanations that are simple but technically accurate
  
• Balancing game design, agent logic, and DataHub integration within hackathon time
  
• Compressing the full experience into a video under 3 minutes
  

Accomplishments that we're proud of

• Creating a working agentic game connected to DataHub
  
• Turning metadata literacy into an interactive experience
  
• Building a smooth loop between gameplay and real metadata updates
  
• Designing a clean, simple visual world that supports learning
  
• Producing a polished demo and documentation within the hackathon timeline
  

What we learned

This project deepened my understanding of:

• How DataHub’s Context Platform enables agentic workflows
  
• How metadata concepts can be taught through interaction
  
• How to design educational experiences that feel natural and fun
  
• How to integrate MCP agents into real‑time applications
  
• How to balance technical depth with simplicity for new learners
  

It also reinforced how powerful agentic systems can be when combined with gamification.

What's next for datahub-data-literacy-game

Next steps include:

• Adding more metadata concepts (tags, glossary terms, schema fields)
  
• Expanding the game world with new levels and puzzles
  
• Introducing quests tied to real DataHub operations
  
• Adding voice‑based agent interactions
  
• Publishing a tutorial so teams can build their own DataHub literacy games
  
• Exploring multiplayer or team‑based metadata challenges
  

The long‑term vision is to turn this into a full DataHub Literacy Playground — a fun, interactive way for teams to learn metadata concepts together.

Built With

• canvas-api
• datahub
• deepseek-api
• fastapi
• mcp-server-datahub
• openai-sdk
• react
• streamlit
• streamlit-frontend:-typescript
• tailwind-css
• typescript
• uvicorn
• vite
• web-audio-api