Work with wizards

Use wizards when you need to create guided, multi-step interactive workflows that collect user input and generate structured output.

Prerequisites

Steps

  1. Choose your wizard approach

    Determine whether to use a built-in wizard, extend an existing one, or create a custom wizard:

    • For debugging workflows, use DebugWizard
    • For code refactoring, use RefactorWizard
    • For security audits, use SecurityWizard
    • For test generation, use TestGenWizard
    • For release preparation, use ReleasePrepWizard

  2. Register or retrieve your wizard

    For built-in wizards, retrieve them with:

    from attune.wizards import get_wizard
wizard_class = get_wizard("debug")  # or "refactor", "security", etc.

    For custom wizards, register them first:

    from attune.wizards import register_wizard
register_wizard("my-custom-wizard", MyCustomWizardClass)

  3. Configure wizard steps

    Create WizardStep instances with the required fields:

    from attune.wizards import WizardStep, StepType

step = WizardStep(
    id="analyze-code",
    name="Code Analysis",
    description="Analyze the codebase for issues",
    step_type=StepType.QUESTION,
    prompt_template="What specific issues should I look for?",
    tier="capable"
)

  4. Initialize and run the wizard

    Create a wizard instance and execute it:

    wizard = wizard_class(ask_user_callback=your_callback_function)
result = wizard.run(initial_context={"project_path": "/path/to/project"})

  5. Process the wizard result

    Handle the WizardResult object returned:

    if result.success:
    print(f"Wizard completed {len(result.steps_completed)} steps")
    print(f"Generated output: {result.generated_output}")
    print(f"Total cost: ${result.total_cost:.2f}")
else:
    print(f"Wizard failed: {result.error}")

  6. Save custom wizard definitions (optional)

    For reusable custom wizards, save the configuration:

    from attune.wizards import save_custom_wizard

wizard_data = {
    "wizard_id": "my-workflow",
    "name": "My Custom Workflow",
    "description": "Custom workflow description",
    "steps": [/* step definitions */]
}
save_custom_wizard(wizard_data)

Verification

Run your wizard and verify:

Key files