Getting Started

LangChain4j CDI is a CDI (Contexts and Dependency Injection) extension that integrates the LangChain4j AI framework with Jakarta EE and Eclipse MicroProfile applications. It enables developers to inject AI services into CDI-managed beans with enterprise features like fault tolerance, telemetry, and external configuration.

What You Can Build

• AI Services -- Declare AI service interfaces and inject them anywhere in your CDI application.
• Agent Orchestration -- Wire multi-agent topologies (sequence, loop, parallel, supervisor, etc.) using CDI annotations.
• MCP Servers -- Expose CDI beans as Model Context Protocol servers with tools, prompts, and resources.
• Enterprise AI -- Add fault tolerance, telemetry, and external configuration to your AI services.

Prerequisites

• Java 17+ (Java 21+ recommended for building)
• Maven 3.8+
• A CDI container (Jakarta EE server or CDI-capable framework)

Quick Start

1. Add the dependency

Choose the extension that matches your runtime:

For Quarkus / Helidon (build-time extension):

<dependency>
    <groupId>dev.langchain4j.cdi</groupId>
    <artifactId>langchain4j-cdi-build-compatible-ext</artifactId>
    <version>${langchain4j-cdi.version}</version>
</dependency>

For WildFly / GlassFish / Liberty / Payara (portable extension):

<dependency>
    <groupId>dev.langchain4j.cdi</groupId>
    <artifactId>langchain4j-cdi-portable-ext</artifactId>
    <version>${langchain4j-cdi.version}</version>
</dependency>

2. Define an AI Service

@RegisterAIService(chatModelName = "#default")
public interface ChatBot {
    String chat(String userMessage);
}

3. Inject and Use

@ApplicationScoped
public class MyBean {

    @Inject
    ChatBot chatBot;

    public String askQuestion(String question) {
        return chatBot.chat(question);
    }
}

Guides

Guide	Description
AI Services	Declare and configure AI service interfaces
Agents	Wire multi-agent topologies via CDI annotations
MCP Server	Turn CDI beans into MCP servers
Expression Language	Dynamic attribute resolution via Jakarta EL and MicroProfile Config
Configuration	External configuration via MicroProfile Config
Fault Tolerance	Retry, timeout, circuit breaker, fallback
Telemetry	OpenTelemetry observability for AI calls
Examples	Sample applications for various runtimes