Embabel Agent Documentation/com.embabel.agent.prompt.persona

Package-level declarations

Types

open class Actor<T> constructor(val persona: T, val llm: <Error class: unknown class>, val toolGroups: Set<String> = emptySet())

An Actor plays a role: Convenient way to combine a PromptContributor with an LLM and tools. Open to allow subclasses to add tools via @Tool methods

CoStar

data class CoStar(val context: String, val objective: String, val style: String, val tone: String, val audience: String, val response: String = "Markdown", separator: String = "#".repeat(12))

CO-STAR prompt framework See https://towardsdatascience.com/how-i-won-singapores-gpt-4-prompt-engineering-competition-34c195a93d41/

Instruction

data class Instruction(val instruction: String)

Simple instruction as persona.

Persona

data class Persona(val name: String, val persona: String, val voice: String, val objective: String)

A way to structure LLM responses, by grounding them in a personality.

RoleGoalBackstory

data class RoleGoalBackstory(val role: String, val goal: String, val backstory: String)

CrewAI style backstory prompt. Included for users migrating from CrewAI. In Embabel, such structures aren't core to the framework, but merely a PromptContributor that can be used in any action implementation.