What causes "validator_name must be provided" in mlflow/mlflow?

Pass validator_name explicitly to the constructor

mlflow/mlflow · error · ValueError

validator_name must be provided

Error message

validator_name must be provided

What it means

The Guardrails scorer base class resolves the validator name either from the explicit `validator_name` argument or from a `validator_name` class variable on the subclass. If both are absent, `__init__` raises this plain ValueError because it cannot determine which Guardrails validator to instantiate.

Source

Thrown at mlflow/genai/scorers/guardrails/__init__.py:74

    Args:
        validator_name: Name of the Guardrails AI validator
        **validator_kwargs: Additional arguments passed to the validator
    """

    _guard: Any = PrivateAttr()

    def __init__(
        self,
        validator_name: str | None = None,
        **validator_kwargs: Any,
    ):
        check_guardrails_installed()

        # Get validator name from class variable if not provided
        if validator_name is None:
            validator_name = getattr(self.__class__, "validator_name", None)
            if validator_name is None:
                raise ValueError("validator_name must be provided")

        super().__init__(name=validator_name)

        from guardrails import Guard, OnFailAction

        validator_class = get_validator_class(validator_name)
        validator = validator_class(on_fail=OnFailAction.NOOP, **validator_kwargs)
        try:
            self._guard = Guard().use(validator)
        except TypeError:
            # guardrails-ai < 0.9.0: on_fail is passed to Guard.use() instead
            self._guard = Guard().use(
                validator_class, on_fail=OnFailAction.NOOP, **validator_kwargs
            )

    def __call__(
        self,
        *,

View on GitHub (pinned to 6a27f2decc)

Solutions

Pass validator_name explicitly to the constructor
Set `validator_name = "YourValidator"` as a class attribute on the subclass
Ensure the subclass is actually instantiated (not the abstract base directly) with a name

Example fix

// before
class MyScorer(GuardrailsScorer):
    pass
// after
class MyScorer(GuardrailsScorer):
    validator_name = "RegexCheck"

Defensive patterns

Strategy: validation

Validate before calling

if validator_name is None and not hasattr(cls, "validator_name"):
    raise ValueError("Subclass must define validator_name or pass it to __init__")

Type guard

def is_named_validator(cls) -> bool:
    return getattr(cls, "validator_name", None) is not None

Try / catch

try:
    scorer = MyGuardrailsScorer()
except ValueError as e:
    if "validator_name must be provided" in str(e):
        scorer = MyGuardrailsScorer(validator_name="RegexCheck")
    else:
        raise

Prevention

Always declare validator_name as a class attribute in custom subclasses
Pass validator_name explicitly at construction instead of relying on inheritance
Lint custom scorer subclasses for the validator_name attribute

When it happens

Trigger: Subclassing the Guardrails scorer base without passing `validator_name` to `__init__` and without declaring `validator_name` as a class attribute; passing an explicit None.

Common situations: Writing a custom validator subclass and forgetting the class variable; refactoring a subclass away from positional args; copying a subclass skeleton that left validator_name unset.

Understand the failure class

Background: "missing required argument" and "the following required arguments were not provided": what required-argument errors mean and how to fix them — this error's family across 20 libraries.

Related errors

AI-assisted analysis of mlflow/mlflow@6a27f2decc (2026-08-29). Data as JSON: /api/errors/32b32e3b635ce34b. Report an issue: GitHub.