Embabel Agent Documentation/com.embabel.agent.api.common/PromptRunner

PromptRunner

interface PromptRunner : LlmUse, PromptRunnerOperations

User code should always use this interface to execute prompts. Typically obtained from an OperationContext or ActionContext parameter, via OperationContext.ai A PromptRunner is immutable once constructed, and has determined LLM and hyperparameters. Use the "with" methods to evolve the state to your desired configuration before executing createObject, generateText or other LLM invocation methods. Thus, a PromptRunner can be reused within an action implementation. A contextual facade to LlmOperations.

See also

Inheritors

FakePromptRunner

Properties

generateExamples

abstract val generateExamples: Boolean?

Whether to generate examples for the prompt. Defaults to unknown: Set to false if generating your own examples.

abstract val llm: <Error class: unknown class>?

abstract val toolGroups: Set<ToolGroupRequirement>

Tool groups exposed. This will include directly registered tool groups and tool groups resolved from ToolGroups.

abstract val toolObjects: List<ToolObject>

Additional objects with @Tool annotation for use in this PromptRunner

Functions

infix inline fun <T> PromptRunner.create(prompt: String): T

Create an object of the given type. Method overloading is evil

open fun <T> createObject(prompt: String, outputClass: Class<T>): T

Create an object of the given type using the given prompt and LLM options from context (process context or implementing class). Prompts are typically created within the scope of an

abstract fun <T> createObject(messages: List<Message>, outputClass: Class<T>): T

Create an object from messages

infix inline fun <T> PromptRunner.createObject(prompt: String): T

Create an object of the given type

createObjectIfPossible

abstract fun <T> createObjectIfPossible(prompt: String, outputClass: Class<T>): T?

Try to create an object of the given type using the given prompt and LLM options from context (process context or implementing class). Prompt is typically created within the scope of an

createObjectIfPossible

inline fun <T> PromptRunner.createObjectIfPossible(prompt: String): T?

abstract fun <T> creating(outputClass: Class<T>): ObjectCreator<T>

Create an object creator for the given output class. Allows setting strongly typed examples.

evaluateCondition

abstract fun evaluateCondition(condition: String, context: String, confidenceThreshold: <Error class: unknown class> = 0.8): Boolean

open infix fun generateText(prompt: String): String

Generate text

open fun respond(messages: List<Message>): AssistantMessage

Respond in a conversation

withContextualPromptContributor

open fun withContextualPromptContributor(contextualPromptContributor: ContextualPromptElement): PromptRunner

withContextualPromptContributors

abstract fun withContextualPromptContributors(contextualPromptContributors: List<ContextualPromptElement>): PromptRunner

Add a prompt contributor that can see context

withGenerateExamples

abstract fun withGenerateExamples(generateExamples: Boolean): PromptRunner

Set whether to generate examples of the output in the prompt on a per-PromptRunner basis. This overrides platform defaults.

@ApiStatus.Experimental

abstract fun withHandoffs(vararg outputTypes: Class<*>): PromptRunner

Add a list of handoffs to agents on this platform

abstract fun withLlm(llm: <Error class: unknown class>): PromptRunner

Specify an LLM for the PromptRunner

withPromptContributor

open fun withPromptContributor(promptContributor: <Error class: unknown class>): PromptRunner

Add a prompt contributor that can add to the prompt. Facilitates reuse of prompt elements.

withPromptContributors

abstract fun withPromptContributors(promptContributors: List<<Error class: unknown class>>): PromptRunner

withPromptElements

open fun withPromptElements(vararg elements: <Error class: unknown class>): PromptRunner

Add varargs of prompt contributors and contextual prompt elements.

open fun withReference(reference: LlmReference): PromptRunner

Add a reference which provides tools and prompt contribution.

open fun withReferences(vararg references: LlmReference): PromptRunner

Add varargs of references which provide tools and prompt contributions.

open fun withReferences(references: List<LlmReference>): PromptRunner

Add a list of references which provide tools and prompt contributions.

@ApiStatus.Experimental

abstract fun withSubagents(vararg subagents: Subagent): PromptRunner

Add a list of subagents to hand off to.

withSystemPrompt

open fun withSystemPrompt(systemPrompt: String): PromptRunner

Add a system prompt

abstract fun withTemplate(templateName: String): TemplateOperations

Use operations from a given template

abstract fun withToolGroup(toolGroup: ToolGroup): PromptRunner

Allows for dynamic tool groups to be added to the PromptRunner.

abstract fun withToolGroup(toolGroup: ToolGroupRequirement): PromptRunner

open fun withToolGroup(toolGroup: String): PromptRunner

Add a tool group to the PromptRunner

open fun withToolGroups(toolGroups: Set<String>): PromptRunner

abstract fun withToolObject(toolObject: ToolObject): PromptRunner

Add a tool object

open fun withToolObject(toolObject: Any?): PromptRunner

Add a tool object to the prompt runner.

withToolObjects

open fun withToolObjects(vararg toolObjects: Any?): PromptRunner

open fun withTools(vararg toolGroups: String): PromptRunner

Add a set of tool groups to the PromptRunner