Decorators API Reference

Decorators for AI testing.

mark.criteria()

def criteria(criteria: str, threshold: float = 7.0)

Mark test with evaluation criteria.

Example:

from testagent import mark

@mark.criteria("output is helpful")
def test_helpfulness():
    return "Hello!"

---

mark.accuracy()

def accuracy(expected: str, threshold: float = 7.0)

Mark test for accuracy comparison.

Example:

@mark.accuracy(expected="4")
def test_math():
    return "4"

---

mark.skip()

def skip(reason: str = "")

Skip a test.

Example:

@mark.skip(reason="Not implemented")
def test_future():
    pass

---

mark.skipif()

def skipif(condition: bool, *, reason: str)

Conditional skip.

Example:

@mark.skipif(sys.platform == 'win32', reason="Unix only")
def test_unix():
    pass

---

mark.xfail()

def xfail(reason: str = "", *, strict: bool = False)

Expected failure.

Example:

@mark.xfail(reason="Known bug")
def test_bug():
    assert False

---

mark.parametrize()

def parametrize(argnames: str, argvalues: list)

Data-driven testing.

Example:

@mark.parametrize("x, y", [(1, 2), (3, 4)])
def test_add(x, y):
    return x + y

---

param()

def param(*values, marks=(), id: str | None = None) -> ParameterSet

Create parameter set with ID.

Example:

@mark.parametrize("x", [
    param(1, id="one"),
    param(2, id="two"),
])
def test_func(x):
    pass

---

fixture()

def fixture(func: Callable = None, *, scope: str = "function")

Create a test fixture.

Scopes: function, class, module, session

Example:

from testagent import fixture

@fixture
def client():
    return Client()

@fixture(scope="session")
def database():
    db = create_db()
    yield db
    db.close()