Implementation:Truera Trulens Block Input Decorator

Knowledge Sources	TruLens TruLens Guardrails
Domains	Guardrails, Safety
Last Updated	2026-02-14 08:00 GMT

Overview

Concrete tool for blocking user inputs that fail a safety or quality check at runtime, provided by the trulens-core library.

Description

The block_input class is a Python decorator that evaluates user input before the decorated method executes. If the input scores below the threshold (or above for inverse metrics), the method returns a safe default value without processing the input.

Usage

Import and apply @block_input to methods that receive user input. Pass a Feedback/Metric instance, threshold, and optional safe return value.

Code Reference

Source Location

Repository: trulens
File: src/core/trulens/core/guardrails/base.py
Lines: L108-196

Signature

class block_input:
    def __init__(
        self,
        feedback: core_metric.Metric,
        threshold: float,
        keyword_for_prompt: Optional[str] = None,
        return_value: Optional[str] = None,
    ):
        """
        Args:
            feedback: Metric instance for input evaluation.
            threshold: Minimum score to allow processing.
            keyword_for_prompt: Kwarg name for the input. Auto-detected if not set.
            return_value: Value to return if input is blocked. Defaults to None.
        """

Import

from trulens.core.guardrails.base import block_input

I/O Contract

Inputs

Name	Type	Required	Description
feedback	Metric	Yes	Feedback function returning a float score
threshold	float	Yes	Minimum score to allow input processing
keyword_for_prompt	str	No	Kwarg name for user input (auto-detected)
return_value	str	No	Safe response returned when input is blocked

Outputs

Name	Type	Description
(decorator)	Callable	Wrapped function that blocks low-scoring inputs

Usage Examples

Block Criminal/Unsafe Inputs

from trulens.core.guardrails.base import block_input
from trulens.core import Feedback
from trulens.providers.openai import OpenAI

provider = OpenAI()
f_criminality = Feedback(provider.criminality, higher_is_better=False)

class SafeApp:
    @block_input(
        feedback=f_criminality,
        threshold=0.9,
        keyword_for_prompt="question",
        return_value="I'm unable to respond to that request."
    )
    def respond(self, question: str) -> str:
        return self.llm.generate(question)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment