pytest-assay 0.1.0

Evaluation framework for Pydantic AI agents

pytest-assay is a framework for the evaluation of Pydantic AI agents. By adding the @pytest.mark.assay decorator to a test, you can run an assay resulting in a readout report which contains the evaluation. The assay compares the current agent responses against pre-recorded baseline responses, e.g. from the main branch. The implementation is using pytest hooks which capture Agent.run() responses inside the test. Below is a minimal example which evaluates the creativity of a search query generation. For a fully functional example see tests/test_plugin_integration.py.

def generate_evaluation_cases():
    ...
    return Dataset(cases)


@pytest.mark.assay(
    generator=generate_evaluation_cases,
    evaluator=PairwiseEvaluator(
        model="openai:gpt-4o",
        criterion="Which of the two search queries shows more genuine curiosity and creativity?",
    ),
)
@pytest.mark.asyncio
async def test_query_generation(context: AssayContext):

    agent = Agent(
        model="openai:gpt-4o",
        system_prompt="Generate a concise web search query for the given research topic.",
    )

    for case in context.dataset.cases:
        async with agent:
            result = await agent.run(user_prompt=f"Generate a search query for the following topic: <TOPIC>{case.inputs['topic']}</TOPIC>")

Both baseline responses and the final readout report are stored in the pytest-assay/ subfolder.

tests
├── pytest-assay
│   └── test_evaluation
│       ├── test_query_generation.json            # Baseline data
│       └── test_query_generation.readout.json    # Readout report
└── test_evaluation.py

The baseline data are generated in new_baseline mode, and the readout report is generated in default evaluate mode.

uv run pytest tests/test_evaluation.py --assay-mode=new_baseline  # Generates baseline data test_query_generation.json
uv run pytest tests/test_evaluation.py --assay-mode=evaluate      # Generates readout report test_query_generation.readout.json