My Agent Framework

A minimal Java agentic AI framework for educational purposes.

Requirements

• Java 21+
• Maven 3.8+
• OpenAI API key (in .env file or environment variable)

Quick Start

# Create .env file with your API key
echo "OPENAI_API_KEY=sk-..." > .env

# Run an example
mvn exec:java -Dexec.mainClass="com.myagent.examples.WeatherAgentExample"

Examples

# Weather Agent - tool calling
mvn exec:java -Dexec.mainClass="com.myagent.examples.WeatherAgentExample"

# Travel Planner - multi-agent workflow
mvn exec:java -Dexec.mainClass="com.myagent.examples.TravelPlannerExample"

# Product Indexer - create vector index
mvn exec:java -Dexec.mainClass="com.myagent.examples.ProductIndexerExample"

# Product Q&A - RAG queries (run indexer first)
mvn exec:java -Dexec.mainClass="com.myagent.examples.ProductQAExample"

# Shopping Assistant - persistent memory
mvn exec:java -Dexec.mainClass="com.myagent.examples.ShoppingAssistantExample"

Admin Dashboard

mvn spring-boot:run

Visit http://localhost:8080/admin/ to view request logs.

The dashboard displays:

• Token usage stats (prompt, completion, total)
• Request logs grouped by Run ID
• Full request/response JSON for each call
• Navigation between log entries

Key Components

Package	Description
core/	Runnable interface, Chain composition
llm/	ChatModel, ChatMessage, OpenAI integration
prompts/	PromptTemplate, ChatPromptTemplate
tools/	@Tool annotation, ToolExecutor
agents/	ChatAgent (ReAct pattern)
memory/	ConversationBufferMemory, FilePersistentMemory
retrieval/	VectorStore, Retriever, TextSplitter

Usage

// Create agent with tools
ChatAgent agent = ChatAgent.builder()
    .chatModel(new OpenAIChatModel(apiKey, "gpt-4o-mini"))
    .toolExecutor(toolExecutor)
    .agentName("MyAgent")
    .build();

String response = agent.invoke("What's the weather?");

License

MIT