Package com.tnsai.planning.tasklist

Class TaskListExecutor

java.lang.Object
com.tnsai.planning.tasklist.TaskListExecutor
public final class TaskListExecutor extends Object
Executes a task list using iterative refinement for each task.

TaskListExecutor is the orchestrator of the Ralph Pattern, combining:

  • Task List Driven - Processes tasks sequentially from a TaskList
  • Clean Context - Creates a fresh RefinementLoop per task
  • Progress Memory - Injects prior learnings via ProgressMemory

Usage


 TaskList tasks = PRDParser.parse(Path.of("prd.json"));

 TaskListResult result = TaskListExecutor.builder()
     .taskList(tasks)
     .progressMemory(ProgressMemory.inMemory())
     .maxIterationsPerTask(5)
     .timeoutPerTask(Duration.ofMinutes(10))
     .onTaskComplete(tr -> log.info("Done: {}", tr.taskItem().name()))
     .build()
     .execute(agent);

 System.out.printf("Completed %d/%d tasks (%.0f%%)%n",
     result.completedCount(), result.totalCount(),
     result.completionRate() * 100);
Author:
TnsAI Team
See Also:

  • Method Details

    • builder

      public static TaskListExecutor.Builder builder()
      Creates a new builder.

    • execute

      public TaskListResult execute(com.tnsai.agents.Agent agent)
      Executes the task list with an agent.
      Parameters:
      agent - the agent to use for refinement
      Returns:
      the execution result

    • execute

      public TaskListResult execute(com.tnsai.llm.LLMClient llm)
      Executes the task list with an LLM client.
      Parameters:
      llm - the LLM client to use
      Returns:
      the execution result