StudyBuddy CLI

StudyBuddy CLI demonstrates the power of building LLM agents with Java instead of Python, using the Embabel framework to create type-safe, production-ready AI applications on the JVM.

Documentación en Español | English Documentation

Why Java for LLM Agents?

This project demonstrates that Java with the Embabel framework is an excellent choice for building LLM agents, offering significant advantages for production AI applications:

Key Advantages:

• Type Safety: Compile-time checking ensures robust code
• Performance: JVM optimization and true multithreading capabilities
• Production Ready: Enterprise-grade monitoring, logging, and deployment
• Dependency Management: Maven's mature dependency resolution
• IDE Support: Excellent refactoring and code intelligence
• Embabel Framework: Purpose-built for Java LLM agents with Spring Boot integration

What Makes This Approach Great:

• Clean Code: Type-safe interfaces and clear contracts
• Maintainability: Strong typing makes refactoring safe and easy
• Enterprise Ready: Seamlessly integrates with existing Java infrastructure
• Spring Ecosystem: Leverage Spring Boot's powerful features

Features

• Powered by Claude 3.5 Haiku: Fast and cost-effective responses
• Bilingual: Native support for Spanish (default) and English
• Three modes:
  • Summarize: Topic summaries with difficulty levels
  • Research: Academic research with configurable depth
  • Explain: Concept explanations with customizable format
• Optimized CLI: Fast interface using PicoCLI
• JVM-based: Built with Embabel framework (Java 21)

Quick Start

Prerequisites

• Java 21+
• Maven 3.6+
• Anthropic API key

Setup

# 1. Clone the project
git clone <repository-url>
cd studybuddy-cli

# 2. Configure environment variables
echo "ANTHROPIC_API_KEY=your_key_here" > .env

# 3. Build the application
./mvnw clean package -DskipTests

# 4. Test functionality
./studybuddy.sh --help

Usage

Available Commands

1. Summarize - Topic Summaries

# Basic summary in Spanish (default)
./studybuddy.sh summarize "machine learning"

# Intermediate summary in English
./studybuddy.sh summarize "photosynthesis" --lang EN --level INTERMEDIATE

# Advanced summary
./studybuddy.sh summarize "quantum computing" --lang EN --level ADVANCED

Available levels: BEGINNER, INTERMEDIATE, ADVANCED

2. Research - Academic Research

# Basic research
./studybuddy.sh research "artificial intelligence" --lang EN --depth BASIC

# Detailed research
./studybuddy.sh research "blockchain technology" --lang EN --depth DETAILED

Available depths: BASIC, DETAILED

3. Explain - Concept Explanations

# Explanation in plain text
./studybuddy.sh explain "gravitation" --lang EN --format PLAIN

# Explanation in Markdown
./studybuddy.sh explain "photosynthesis" --lang EN --format MARKDOWN

Available formats: PLAIN, MARKDOWN

Global Options

• --lang [ES|EN]: Output language (default: Spanish)
• --help: Show command help

Architecture

Main Components

StudyBuddy-CLI/
├── StudyBuddyAgent          # Main agent with 3 actions
│   ├── summarizeTopic()     # Educational summaries
│   ├── researchTopic()      # Academic research  
│   └── explainConcept()     # Detailed explanations
├── LocalizationService      # i18n system
├── CLI Commands             # PicoCLI interface
└── Domain Objects           # Type-safe data model

Technologies

• Framework: Embabel Agent Framework 0.1.2
• LLM: Claude 3.5 Haiku Latest
• Templates: Freemarker for localization
• CLI: PicoCLI + Spring Boot
• Build: Maven + Java 21

Model Configuration

# Claude 3.5 Haiku optimized for education
embabel.models.defaultLlm=claude-3-5-haiku-latest
embabel.ai.anthropic.temperature=0.3
embabel.ai.anthropic.max-tokens=2000

Estimated Costs

With Claude 3.5 Haiku:

• Input: $0.80 per 1M tokens
• Output: $4 per 1M tokens
• Prompt Caching: 90% discount
• Typical cost per request: ~$0.007 (less than 1 cent)

Development

Project Structure

src/main/java/com/michead/studybuddy/
├── agent/
│   └── StudyBuddyAgent.java
├── domain/
│   ├── StudyRequest.java
│   ├── ResearchRequest.java
│   ├── ExplanationRequest.java
│   ├── StudySummary.java
│   ├── ResearchResult.java
│   └── ConceptExplanation.java
├── localization/
│   ├── StudyBuddyLocale.java
│   ├── LocalizationService.java
│   └── LocalizationConfig.java
└── cli/
    ├── StudyBuddyCLI.java
    ├── SummarizeCommand.java
    ├── ResearchCommand.java
    └── ExplainCommand.java

Localization Templates

src/main/resources/templates/prompts/
├── es/
│   ├── research.ftl
│   ├── educational_content.ftl
│   └── summarization.ftl
└── en/
    ├── research.ftl
    ├── educational_content.ftl
    └── summarization.ftl

Running Tests

# Run all tests
./mvnw test

# Specific localization test
./mvnw test -Dtest=LocalizationServiceTest

JVM Advantages Demonstration

This project demonstrates JVM advantages for LLM agents mentioned in Rod Johnson's article:

• Type Safety: No magic strings, everything type-safe
• Less Code: More concise than Python equivalents
• Production Ready: Logging, configuration, concurrency handling
• Better Integration: Seamless with Java/Spring ecosystem
• Performance: JVM optimizations + Claude 3.5 Haiku speed

StudyBuddy-CLI is the perfect proof-of-concept for LLM agent potential on the JVM.

Roadmap

• Support for more languages (French, Italian)
• Integration with academic databases
• Export to PDF/Word
• Complementary Web UI
• IDE plugins
• GraalVM native image support

Contributing

1. Fork the project
2. Create feature branch (git checkout -b feature/new-feature)
3. Commit changes (git commit -am 'Add new feature')
4. Push branch (git push origin feature/new-feature)
5. Open Pull Request

License

Licensed under the Apache License, Version 2.0

Support

If you encounter issues or have questions:

1. Check the Spanish documentation
2. Verify your API key configuration
3. Run with --help to see all available options

---

StudyBuddy-CLI - Demonstrating the power of LLM agents on the JVM with Claude 3.5 Haiku