Embabel Agent Documentation/com.embabel.chat.agent

Package-level declarations

Types

class AgentPlatformChatSession(val user: User?, val outputChannel: OutputChannel, planLister: PlanLister, val processOptions: ProcessOptions = ProcessOptions(), val responseGenerator: ResponseGenerator, val conversation: Conversation = InMemoryConversation()) : ChatSession

Support for chat sessions leveraging an AgentPlatform.

AgentProcessChatbot

class AgentProcessChatbot(agentPlatform: AgentPlatform, agentSource: AgentSource, listenerProvider: ListenerProvider = ListenerProvider { user, outputChannel -> emptyList() }) : Chatbot

Chatbot implementation backed by an AgentProcess The AgentProcess must react to messages and respond on its output channel The AgentProcess can assume that the Conversation will be available in the blackboard, and the latest UserMessage. Action methods will often take precondition being that the last event was a UserMessage. A convenient approach is for the core action methods to return ChatbotReturn, and handle ConversationOver, although that is not required.

AgentResponseGenerator

class AgentResponseGenerator(agentPlatform: AgentPlatform, agent: Agent) : ResponseGenerator

Respond to user messages using an agent.

AgentSource

fun interface AgentSource

AutonomyResponseGenerator

class AutonomyResponseGenerator(autonomy: Autonomy, goalChoiceApprover: GoalChoiceApprover, val processWaitingHandler: ProcessWaitingHandler, val chatConfig: ChatConfig) : ResponseGenerator

Respond to messages by choosing and executing goals using Autonomy. Based on last user input.

BlackboardEntryFormatter

interface BlackboardEntryFormatter

BlackboardFormatter

interface BlackboardFormatter

ChatbotReturn

data class ChatbotReturn(val assistantMessage: AssistantMessage? = null, val conversationTermination: ConversationTermination? = null) : SomeOf

Return from a chatbot agent action method

ChatConfig

data class ChatConfig(val confirmGoals: Boolean = true, val bindConversation: Boolean = false, val multiGoal: Boolean = false, val model: String = OpenAiModels.GPT_41_MINI, val temperature: Double? = null)

Configuration for the chat session

ConversationTermination

data class ConversationTermination(val reason: String)

DefaultBlackboardEntryFormatter

object DefaultBlackboardEntryFormatter : BlackboardEntryFormatter

DefaultBlackboardFormatter

class DefaultBlackboardFormatter(entryFormatter: BlackboardEntryFormatter = DefaultBlackboardEntryFormatter) : BlackboardFormatter

DefaultChatAgentBuilder

class DefaultChatAgentBuilder(autonomy: Autonomy, llm: <Error class: unknown class>, persona: Persona = K9, promptTemplate: String = "chat/default_chat", blackboardFormatter: BlackboardFormatter = DefaultBlackboardFormatter())

ListenerProvider

fun interface ListenerProvider

ProcessWaitingHandler

interface ProcessWaitingHandler

Handles process waiting exceptions in a platform-specific way

ResponseGenerator

interface ResponseGenerator

Generates response(s) in a chat session.

Properties

val K9: Persona

SLASH_COMMAND_REGEX

val SLASH_COMMAND_REGEX: <Error class: unknown class>

Functions

parseSlashCommand

fun parseSlashCommand(input: String): <Error class: unknown class><String, List<String>>?