Class GoalOrientedPlanner

java.lang.Object

dev.langchain4j.agentic.patterns.goap.GoalOrientedPlanner

All Implemented Interfaces:

Planner

public class GoalOrientedPlanner
extends Object
implements Planner

Constructor Summary

Constructors

Constructor	Description
GoalOrientedPlanner()

Method Summary

Modifier and Type	Method	Description
Action	firstAction(PlanningContext planningContext)	Returns the first action to execute when the planner starts (or resumes).
void	init(InitPlanningContext initPlanningContext)
Action	nextAction(PlanningContext planningContext)	Determines the next action to execute based on the result of the previous agent invocation.
void	restoreExecutionState(Map<String,Object> state)	GoalOrientedPlanner does not persist execution state because firstAction(PlanningContext) recomputes the path from the current scope state via graph search.
AgenticSystemTopology	topology()	Returns the topology of the agentic system managed by this planner.

Methods inherited from class Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface Planner

as, call, call, done, done, executionState, noOp, suspend, terminated

Constructor Details

GoalOrientedPlanner

public GoalOrientedPlanner()

Method Details

init

public void init(InitPlanningContext initPlanningContext)

Specified by:

init in interface Planner

firstAction

public Action firstAction(PlanningContext planningContext)

Description copied from interface: Planner

Returns the first action to execute when the planner starts (or resumes). Defaults to delegating to Planner.nextAction(PlanningContext).

Specified by:

firstAction in interface Planner

Parameters:

planningContext - the current planning context

Returns:

the first action to execute

nextAction

public Action nextAction(PlanningContext planningContext)

Description copied from interface: Planner

Determines the next action to execute based on the result of the previous agent invocation.

Specified by:

nextAction in interface Planner

Parameters:

planningContext - the current planning context, including the previous agent's result

Returns:

the next action to execute

restoreExecutionState

public void restoreExecutionState(Map<String,Object> state)

GoalOrientedPlanner does not persist execution state because firstAction(PlanningContext) recomputes the path from the current scope state via graph search. On recovery, completed agents' outputs are already in scope, so the search produces a shorter path containing only the remaining agents. The cursor resets to 0 naturally, making state persistence unnecessary and potentially harmful (a stale cursor could point beyond the bounds of the recomputed path).

Specified by:

restoreExecutionState in interface Planner

Parameters:

state - the previously saved execution state

topology

public AgenticSystemTopology topology()

Description copied from interface: Planner

Returns the topology of the agentic system managed by this planner.

Specified by:

topology in interface Planner

Returns:

the topology (defaults to AgenticSystemTopology.SEQUENCE)