Package-level declarations

Types

Link copied to clipboard

A class for Agent Feature that can be added to an agent pipeline, The feature stands for providing specific functionality and configuration capabilities.

Link copied to clipboard

Represents a graph-specific AI agent feature that can be installed into an instance of AIAgentGraphPipeline.

Link copied to clipboard
class AIAgentGraphPipeline(val clock: Clock = Clock.System) : AIAgentPipeline

Represents a pipeline for AI agent graph execution, extending the functionality of AIAgentPipeline. This class manages the execution of specific nodes in the pipeline using registered handlers.

Link copied to clipboard

Represents a non-graph-specific AI agent feature that can be installed into an instance of AIAgentNonGraphPipeline.

Link copied to clipboard
class AIAgentNonGraphPipeline(val clock: Clock = Clock.System) : AIAgentPipeline

Represents a specific implementation of an AI agent pipeline that does not use graph-based processing. This class inherits from the base AIAgentPipeline class and may be used for handling workflows or data processing tasks that do not require graph-based data structures.

Link copied to clipboard
abstract class AIAgentPipeline(val clock: Clock)

Pipeline for AI agent features that provides interception points for various agent lifecycle events.

Link copied to clipboard
data class InterceptContext<TFeature : Any>(val feature: AIAgentFeature<*, TFeature>, val featureImpl: TFeature)

Represents the context for intercepting or interacting with a specific feature in an AI agent pipeline.

Link copied to clipboard

A wrapper around ai.koog.prompt.executor.model.PromptExecutor that allows for adding internal functionality to the executor to catch and log events related to LLM calls.