SLIM Guessing Game - Educational AI Demo

An educational demo showcasing AI Agents and SLIM's secure messaging for children. Watch AI agents play "20 Questions" together!

Purpose: This is a demonstration application created to teach children about AI agents and secure communication using the SLIM Platform.

The Game

AI agents play "20 Questions" - one agent thinks of an object, others ask questions to guess it!

The Players (AI Agents)

1. Coordinator - Game master (manages rules and turns)
2. Alice (Thinker) - Thinks of a secret object and answers yes/no questions
3. Bob, Carol, Dave (Guessers) - Ask questions and try to guess the object
4. Travis (Observer) - Watches the game and translates messages to other languages

How They Communicate

┌─────────────────────────────────────────────────────────────────────┐
│                   PUBLIC GROUP SESSION (Broadcast)                  │
│                                                                     │
│  Coordinator ◄──► Alice ◄──► Bob ◄──► Carol ◄──► Dave ◄──► Travis   │
│                                                                     │
│  All game communication: questions, answers, guesses, game state    │
│  Everyone sees everything on this channel                           │
└─────────────────────────────────────────────────────────────────────┘


┌─────────────────────────────────────────────────────────────────────┐
│              SECRET 1:1 SESSION (Private, Encrypted)                │
│                                                                     │
│              Alice ──────────────────────► Travis                   │
│                     "Crayon" (secret)                               │
│                                                                     │
│  Only Alice and Travis participate in this channel                  │
│  MLS encrypted - Coordinator, Bob, Carol, Dave cannot see it        │
│  Used for observability/monitoring without affecting game           │
└─────────────────────────────────────────────────────────────────────┘

Two channels running at the same time:

• Public channel: Everyone talks about the game
• Secret channel: Alice secretly tells Travis the answer (for monitoring)

Quick Start

Want to run this? See the Quick Start Guide for detailed setup instructions.

TL;DR:

# 1. Install prerequisites: Docker, Kind, Task, Helm, and an LLM provider
# 2. Configure LLM credentials
cp deployments/.env.example deployments/.env
# Edit deployments/.env with your LLM settings

# 3. Start everything
task setup

# 4. Watch the game
task game:logs

# 5. Stop everything
task teardown

What Children Learn

This demo teaches:

• AI Agents - How AI programs work together
• Secure Communication - How messages can be private or public
• Strategic Thinking - Different ways to solve problems
• Natural Language - How AI understands questions

Useful Commands

See QUICKSTART.md for full commands and troubleshooting.

task game:logs     # Watch the game
task game:status   # Check status
task game:restart  # Restart game
task teardown      # Stop everything

Technical Details

Built with:

• SLIM Platform - Secure agent messaging (MLS encryption)
• SLIM Docs
• Python 3.11 - Agent logic
• Kubernetes/Docker/Kind - Container orchestration
• LLM - AI intelligence (OpenAI, Azure, or local Ollama)

Demonstrates:

• Dual-session architecture - Public broadcast + private 1:1 channels
• MLS encryption - End-to-end secure messaging
• AI agent collaboration - Multiple agents working together
• Turn-based coordination - Managing sequential interactions

For detailed implementation, see the src/guessing_game/ source code.

License

Apache 2.0 - see LICENSE file