Implementation:Guardrails ai Guardrails Guard Use Streaming

Knowledge Sources	Guardrails Guardrails Streaming
Domains	Streaming, Validation
Last Updated	2026-02-14 00:00 GMT

Overview

Concrete method for configuring a Guard with validators for streaming validation provided by the guardrails package.

Description

This is the same Guard.use method used for non-streaming validation, but applied in the context of streaming. The validator instance attached via .use() should support streaming through its _chunking_function method. The same fluent API pattern applies: validators are attached with a target (on="output") and registered in the Guard's validator map.

Usage

Use Guard().use(validator) exactly as for non-streaming, but be aware that for streaming calls, validators with string output will use their _chunking_function to determine validation boundaries.

Code Reference

Source Location

• Repository: guardrails
• File: guardrails/guard.py
• Lines: L907-940

Signature

def use(
    self,
    validator: UseValidatorSpec,
    *args,
    on: str = "output",
    **kwargs,
) -> "Guard":
    """Use a validator to validate either of the following:
    - The output of an LLM request
    - The message history

    Args:
        validator: The validator to use. Either the class or an instance.
        on: The part of the LLM request to validate. Defaults to "output".
    """

Import

from guardrails import Guard

I/O Contract

Inputs

Name	Type	Required	Description
validator	UseValidatorSpec	Yes	A Validator instance (should implement _chunking_function for streaming)
on	str	No	Validation target: "output" (default)

Outputs

Name	Type	Description
self	Guard	Returns the Guard instance (fluent chaining)

Usage Examples

Streaming Guard Setup

from guardrails import Guard
from guardrails.hub import ToxicLanguage

# Same .use() API as non-streaming
guard = Guard().use(ToxicLanguage(on_fail="fix"))

# The difference is in how it's called (stream=True)
stream = guard(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Tell me a story."}],
    stream=True,
)
for chunk in stream:
    print(chunk.validated_output, end="", flush=True)

Related Pages

Implements Principle

• Principle:Guardrails_ai_Guardrails_Guard_Streaming_Configuration