Skip to main content

Spring Boot Integration

The Arc libraries are commonly used in Spring Boot applications. For this purpose, the arc-spring-boot-starter library is provided. Add the following dependencies to your project. For example, in the build.gradle.kts file.

implementation("io.github.lmos-ai.arc:arc-scripting:$arcVersion")
implementation("io.github.lmos-ai.arc:arc-azure-client:$arcVersion")
implementation("io.github.lmos-ai.arc:arc-spring-boot-starter:$arcVersion")

The arc-spring-boot-starter library will set up the necessary beans and configurations to run the Arc Agents within a Spring Boot application. Providing many defaults that can be overridden.

The only bean that is required to be defined is the ChatCompleterProvider bean. The ChatCompleterProvider bean provides instances of the ChatCompleter interface.

The ChatCompleter interface is usually backed by an AI service that can complete text, such as Large Language Models. The ChatCompleterProvider enables these instances to be created dynamically based on the current context.

One implementation of the ChatCompleter interface is the AzureAIClient class. This class uses the Azure OpenAI API.

Example of Spring Boot configuration

    @Bean
    fun aiClient(languageModel: LanguageModel, client: OpenAIAsyncClient, eventPublisher: EventPublisher) =
        AzureAIClient(languageModel, client, eventPublisher)

    @Bean
    fun chatCompleterProvider(client: AzureAIClient) = ChatCompleterProvider { client }

Agent Scripting​

The following properties can be set in the application.yml file to configure the Arc Agent Scripting.

arc:
  scripts:
    folder: ./agents // the folder to load the scripts from
    hotReload:
      enable: true
      delay: PT1S