Skip to main content
Recipes package agents, tasks, and tools so you can run repeatable workflows from YAML. 
from praisonaiagents import Agent

agent = Agent(name="Recipe Runner", instructions="Summarise what PraisonAI recipes provide.")
agent.start("What can I do with recipes?")
The user opens a recipe guide, configures roles, and runs praisonai recipe run with variables.

​
Quick Start

1

Simple Usage

List and run a recipe from the CLI.
praisonai recipe list
praisonai recipe run my-recipe
2

With Configuration

Pass variables into the recipe at run time.
praisonai recipe run my-recipe --var topic="AI agents"

​
How It Works

​
Recipe Guides

Learn how to create, use, manage, and debug recipes in PraisonAI.

Create Custom Recipes

Build your own recipes from scratch

Use Existing Recipes

Run and configure existing recipes

Add Tools to Recipes

Configure tools for your recipes

Manage Recipes

Update, edit, and delete recipes

Debug Recipes

Troubleshoot recipe issues

Different Ways to Create

Explore all recipe creation methods

​
Quick Reference

TaskCommand
List recipespraisonai recipe list
Add from GitHubpraisonai recipe add github:user/repo/recipe
Add localpraisonai recipe add ./my-recipe
Run recipepraisonai recipe run <name>
Get infopraisonai recipe info <name>
Validatepraisonai recipe validate <path>
Init newpraisonai recipe init <name>

​
Best Practices

Define roles, goals, tools, and tasks in one agents.yaml. Recipes stay portable when everything lives in the recipe directory.
praisonai recipe validate <path> catches schema errors before execution. Wire it into CI so broken recipes never ship.
Use {{task}}-style placeholders and pass values with --var key=value so one recipe serves many inputs.