Decorators¶

pytest-like decorators for AI tests.

graph LR
    subgraph "Mark Decorators"
        A["@mark.criteria"] --> B[Test with criteria]
        C["@mark.accuracy"] --> D[Test accuracy]
        E["@mark.skip"] --> F[Skip test]
        G["@mark.skipif"] --> H[Conditional skip]
        I["@mark.xfail"] --> J[Expected failure]
    end

    classDef decorator fill:#6366F1,stroke:#7C90A0,color:#fff
    classDef desc fill:#10B981,stroke:#7C90A0,color:#fff

    class A,C,E,G,I decorator
    class B,D,F,H,J desc

@mark.criteria¶

Test output against criteria:

from testagent import mark

@mark.criteria("output is helpful and accurate")
def test_helpfulness():
    return "Hello! I'm here to help you."

Parameters:

Parameter	Type	Default	Description
`criteria`	`str`	-	Criteria to evaluate
`threshold`	`float`	`7.0`	Pass threshold

@mark.accuracy¶

Test output accuracy:

from testagent import mark

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

@mark.skip¶

Skip a test:

from testagent import mark

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

@mark.skipif¶

Conditional skip:

from testagent import mark
import sys

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

@mark.xfail¶

Expected failure:

from testagent import mark

@mark.xfail(reason="Known bug #123")
def test_known_bug():
    assert False

@mark.xfail(reason="Should pass", strict=True)
def test_strict():
    # If this passes, test fails
    pass

Parameters:

Parameter	Type	Default	Description
`reason`	`str`	`""`	Reason for expected failure
`strict`	`bool`	`False`	Fail if test passes