ADK Multi-Agent Educator

A sophisticated multi-agent educational system built with Google's ADK framework, featuring specialized AI tutors for Math, Science, Music, and English. Each agent is expertly trained in their subject domain and works together through an intelligent coordinator to provide comprehensive, personalized educational support.

✨ Features

• 🎯 Four Specialized Agents: Expert tutors for Math, Science, Music, and English
• 🤝 Intelligent Coordination: Smart routing to the most appropriate agent
• ✍️ Grammar & Writing Support: Comprehensive English language assistance
• 📊 ADK Integration: Built on Google's Agent Development Kit
• 🚀 Minimal Dependencies: Only requires google-adk
• 🌐 Web Interface: Easy access through adk web command

🚀 Quick Start

Installation

1. Clone the repository:

git clone <repository-url>
cd adk-multi-agent-educator

2. Install dependencies:

pip install -r requirements.txt

3. Launch ADK Web Interface:

adk web

4. Start learning! Ask questions like:
   • "What is 2+2?" (Math Agent)
   • "How does photosynthesis work?" (Science Agent)
   • "Explain major scales" (Music Agent)
   • "Fix this sentence grammar" (English Agent)

🎓 Agent Capabilities

📐 Math Agent

• Core Topics: Algebra, Geometry, Calculus, Statistics, Arithmetic
• Teaching Style: Step-by-step problem solving with clear explanations
• Special Features: Multiple solution methods, real-world applications
• Example: "How do I solve x² + 5x + 6 = 0?"

🔬 Science Agent

• Core Topics: Physics, Chemistry, Biology, Earth Sciences
• Teaching Style: Scientific method approach with experiments
• Special Features: Virtual experiments, theory-to-practice connections
• Example: "Why do objects fall at the same rate in a vacuum?"

🎵 Music Agent

• Core Topics: Music theory, composition, performance, history
• Teaching Style: Both theoretical and practical approaches
• Special Features: Cultural context, creative exercises
• Example: "What's the difference between major and minor chords?"

✍️ English Agent New!

• Core Topics: Grammar, writing, vocabulary, reading comprehension
• Teaching Style: Clear explanations with examples and practice
• Special Features: Writing feedback, style improvement, language rules
• Example: "What's the difference between 'affect' and 'effect'?"

🏗️ Architecture

src/
├── root_agent.py                    # Main ADK coordinator
└── adk_educator/
    ├── prompt.py                    # Coordination prompts
    └── sub_agents/                  # Specialized agents
        ├── math_agent/
        ├── science_agent/
        ├── music_agent/
        └── english_agent/

The system uses Google's ADK LlmAgent and AgentTool to create a coordinator that intelligently routes student questions to the most appropriate specialist agent.

💡 How It Works

1. Student asks a question through the ADK web interface
2. Educational Coordinator analyzes the question and determines the best agent
3. Specialist Agent provides expert assistance in their domain
4. Response is delivered with educational context and follow-up suggestions

🔧 Technical Details

Requirements

• Google ADK: google-adk>=0.1.0
• Python: 3.8+

Agent Structure

Each specialist agent is built using:

from google.adk.agents import LlmAgent

agent = LlmAgent(
    name="agent_name",
    model="gemini-2.5-pro-preview-05-06",
    description="Agent specialization",
    instruction="Detailed teaching prompt",
    output_key="response_key"
)

Coordinator Pattern

The main coordinator uses AgentTool to access specialists:

educational_coordinator = LlmAgent(
    name="educational_coordinator",
    tools=[
        AgentTool(agent=math_agent),
        AgentTool(agent=science_agent),
        AgentTool(agent=music_agent),
        AgentTool(agent=english_agent),
    ]
)

🎯 Example Interactions

Math Problem Solving

Student: "I need help with calculus derivatives"
Coordinator: Routes to Math Agent
Math Agent: Provides step-by-step derivative rules with examples

Grammar Correction

Student: "Can you fix this sentence: 'Me and my friend went to store'"
Coordinator: Routes to English Agent
English Agent: Corrects to "My friend and I went to the store" with explanation

Science Concepts

Student: "How do solar panels work?"
Coordinator: Routes to Science Agent
Science Agent: Explains photovoltaic effect with practical applications

Music Theory

Student: "What makes a chord progression sound good?"
Coordinator: Routes to Music Agent
Music Agent: Explains harmony, tension, and resolution with examples

🔮 Future Enhancements

• Additional Subjects: History, Art, Computer Science agents
• Multimodal Support: Image analysis, audio processing
• Learning Analytics: Progress tracking and personalized recommendations
• Collaborative Features: Multi-student sessions and peer learning
• Assessment Tools: Quiz generation and automated grading

📄 License

Copyright 2025 Google LLC - Licensed under the Apache License, Version 2.0

🤝 Contributing

This project follows Google's ADK best practices and the academic research patterns for multi-agent educational systems.

---

Ready to start learning? Just run adk web and ask any educational question! 🚀