agents::workflows::PromptChainingWorkflow Class Reference

A workflow that chains multiple prompts together. More...

#include <prompt_chaining_workflow.h>

Inheritance diagram for agents::workflows::PromptChainingWorkflow:

Classes
struct	Step
	Step in the workflow. More...

Public Member Functions
	PromptChainingWorkflow (std::shared_ptr< AgentContext > context)
	Constructor with context.
void	addStep (const Step &step)
	Add a step to the workflow.
void	addStep (const String &name, const String &prompt_template, std::function< bool(const JsonObject &)> validator=nullptr, std::function< JsonObject(const JsonObject &)> transformer=nullptr)
	Add a step to the workflow with basic params.
JsonObject	run (const String &input) override
	Execute the workflow with input.
JsonObject	run ()
	Execute the workflow using the latest USER message from context memory.
virtual void	runAsync (const String &input, std::function< void(const JsonObject &)> callback)
	Run the workflow with a user input asynchronously.
std::shared_ptr< AgentContext >	getContext () const
	Get the workflow's context.
void	setStepCallback (std::function< void(const String &, const JsonObject &)> callback)
	Set a callback for intermediate steps.
void	setMaxSteps (int max_steps)
	Set maximum number of steps.
int	getMaxSteps () const
	Get maximum number of steps.

Detailed Description

A workflow that chains multiple prompts together.

This workflow decomposes a task into a sequence of steps, where each LLM call processes the output of the previous one. It can add programmatic checks between steps to ensure the process is on track.

Examples

actor_agent_example.cpp.

Constructor & Destructor Documentation

PromptChainingWorkflow()

agents::workflows::PromptChainingWorkflow::PromptChainingWorkflow

(

std::shared_ptr< AgentContext >

context

)

Constructor with context.

Parameters

context

The context to use

Member Function Documentation

addStep() [1/2]

void agents::workflows::PromptChainingWorkflow::addStep

(

const Step &

step

)

Add a step to the workflow.

Parameters

step	The step to add

addStep() [2/2]

void agents::workflows::PromptChainingWorkflow::addStep	(	const String &	name,
		const String &	prompt_template,
		std::function< bool(const JsonObject &)>	validator = nullptr,
		std::function< JsonObject(const JsonObject &)>	transformer = nullptr )

Add a step to the workflow with basic params.

Parameters

name	The name of the step
prompt_template	The prompt template for the step
validator	The validator function for the step
transformer	The transformer function for the step

getContext()

std::shared_ptr< AgentContext > agents::workflows::Workflow::getContext ( ) const

inherited

Get the workflow's context.

Returns

The context

getMaxSteps()

int agents::workflows::Workflow::getMaxSteps ( ) const

inherited

Get maximum number of steps.

Returns

The maximum number of steps

run() [1/2]

JsonObject agents::workflows::PromptChainingWorkflow::run

(

)

Execute the workflow using the latest USER message from context memory.

Returns

The result of the workflow execution

run() [2/2]

JsonObject agents::workflows::PromptChainingWorkflow::run ( const String & input )

overridevirtual

Execute the workflow with input.

Parameters

input

The input to execute the workflow with

Returns

The result of the workflow execution

Implements agents::workflows::Workflow.

runAsync()

virtual void agents::workflows::Workflow::runAsync ( const String & input, std::function< void(const JsonObject &)> callback )

virtualinherited

Run the workflow with a user input asynchronously.

Parameters

input	The user input
callback	The callback

setMaxSteps()

void agents::workflows::Workflow::setMaxSteps ( int max_steps )

inherited

Set maximum number of steps.

Parameters

max_steps

The maximum number of steps

setStepCallback()

void agents::workflows::Workflow::setStepCallback ( std::function< void(const String &, const JsonObject &)> callback )

inherited

Set a callback for intermediate steps.

Parameters

callback

The callback

Examples

prompt_chain_example.cpp.