Working with Scenarios

Scenarios let you test the same agent with different configurations and dependencies. This is essential for agents that need to work with different users, environments, or settings.

Example: Personal Assistant Agent

Let's say you have a personal assistant agent that books meetings and leaves messages. The agent needs to know who it's assisting to personalize responses and access their preferences.

Define Your Dependencies

First, create a model for the user information your agent needs:

from pydantic import BaseModel


class UserInfo(BaseModel):
    name: str
    email: str
    timezone: str
    preferences: str

Export with Multiple Scenarios

Now export your agent with different scenarios for different users:

from agent_playbook import export

export(
    agent=assistant,
    agent_name="Personal Assistant",
    scenarios=[
        {
            "name": "Alice (CEO)",
            "description": "Executive assistant for Alice",
            "settings": UserInfo(
                name="Alice Chen",
                email="alice@company.com",
                timezone="America/New_York",
                preferences="Prefers morning meetings, keep responses brief",
            ),
        },
        {
            "name": "Bob (Engineer)",
            "description": "Assistant for Bob the engineer",
            "settings": UserInfo(
                name="Bob Smith",
                email="bob@company.com",
                timezone="America/Los_Angeles",
                preferences="No meetings before 10am, prefers afternoon focus time",
            ),
        },
        {
            "name": "Carol (Sales)",
            "description": "Sales assistant for Carol",
            "settings": UserInfo(
                name="Carol Johnson",
                email="carol@company.com",
                timezone="Europe/London",
                preferences="Works across timezones, prefers quick responses",
            ),
        },
    ],
)

Save this as personal_assistant__scenarios.py in your project.