feat: new Scenario API ✨

[rogeriochaves]

@pytest.mark.agent_test
@pytest.mark.asyncio
async def test_ai_assistant_agent():
    scenario = Scenario(
        name="false assumptions",
        description="""
            The agent makes false assumption about being an ATM bank, and user corrects it
        """,
        agent=AiAssistantAgentAdapter,
        criteria=[
            "user should get good recommendations on river crossing",
            "agent should NOT follow up about ATM recommendation after user has corrected them they are just hiking",
        ],
        max_turns=5,
    )

    result = await scenario.script(
        [
            scenario.user("how do I safely approach a bank?"),
            scenario.agent(),
            scenario.user(),
            scenario.agent(),
            scenario.judge(),
        ]
    ).run()

    assert result.success

[Copilot]

Pull Request Overview

This PR introduces a new Scenario API with improvements in agent adapter configuration, scenario scripting, and enhanced type safety across several modules. Key changes include:

• Conversion of agent functions to subclass-based adapters for both testing and production.
• Refactoring of configuration merging and scenario scripting for better consistency.
• Updates to tests and examples to support the new workflow and API signatures.

Reviewed Changes

Copilot reviewed 26 out of 26 changed files in this pull request and generated 1 comment.

File	Description
tests/*	Updated tests to use new adapter subclass pattern
scenario/*	Refactored Scenario initialization, configuration, and scripting
examples/*	Updated examples to align with new Scenario API design
setup.py, pyproject.toml, Makefile	Version updates and dependency additions

Comments suppressed due to low confidence (2)

scenario/config.py:32

• [nitpick] Instead of creating a custom items() helper to merge configuration values, consider using Pydantic’s model_dump(exclude_none=True) directly to improve clarity and consistency.

def merge(self, other: "ScenarioConfig") -> "ScenarioConfig":

scenario/scenario.py:100

• Filtering None values out of the agents list before type and subclass checks would prevent potential unclear ValueErrors; consider using a list comprehension to include only non-null agents.

agents = agents or [kwargs.get("testing_agent"), agent,  # type: ignore]

[Copilot]

[nitpick] The check for an empty model value may be ambiguous if self.model is an empty string. Consider a more explicit validation or providing a sensible default in TestingAgent.with_config.

Suggested change

	if not self.model:
	if not isinstance(self.model, str) or not self.model.strip():