Step 2: Meal Plan

Generate multiple recipes at once using parallel execution.

Goal

Generate breakfast, lunch, and dinner recipes simultaneously.

The Workflow

Update recipe.yaml:

# yaml-language-server: $schema=../../schemas/workflow.schema.json
name: meal-plan
description: Generate three meals in parallel

inputs:
  - name: diet
    type: string
    description: Dietary preference (e.g., high-protein, vegan, keto)

steps:
  # Using 'fast' model to minimize token usage during tutorials
  - id: meals
    type: parallel
    max_concurrency: 3
    steps:
      - id: breakfast
        type: llm
        model: fast
        prompt: |
          Generate a {{.inputs.diet}} breakfast recipe.
          Include the recipe name, ingredients with quantities, and cooking steps.

      - id: lunch
        type: llm
        model: fast
        prompt: |
          Generate a {{.inputs.diet}} lunch recipe.
          Include the recipe name, ingredients with quantities, and cooking steps.

      - id: dinner
        type: llm
        model: fast
        prompt: |
          Generate a {{.inputs.diet}} dinner recipe.
          Include the recipe name, ingredients with quantities, and cooking steps.

outputs:
  - name: breakfast
    type: string
    value: "{{.steps.meals.breakfast.response}}"
    description: Breakfast recipe
    format: markdown

  - name: lunch
    type: string
    value: "{{.steps.meals.lunch.response}}"
    description: Lunch recipe
    format: markdown

  - name: dinner
    type: string
    value: "{{.steps.meals.dinner.response}}"
    description: Dinner recipe
    format: markdown

Run It

conductor run recipe.yaml
conductor run recipe.yaml -i diet="keto"

All three recipes generate at the same time.

What You Learned

• Parallel - Use type: parallel with nested steps
• max_concurrency - Limit concurrent executions
• Nested outputs - Reference parallel step outputs with {{.steps.parent.child.response}}

Next

Step 3: Pantry Check - Read ingredients from a file.