Implementation:Guardrails ai Guardrails Guard Call Stream

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

Overview

Concrete method for invoking a Guard in streaming mode provided by the guardrails package.

Description

This is Guard.__call__ with stream=True passed as a keyword argument. When stream is True, the _exec method creates a StreamRunner instead of a regular Runner. The StreamRunner yields ValidationOutcome objects for each validated chunk. Note that num_reasks should be 0 for streaming as re-asking is not supported.

Usage

Call the Guard with stream=True and iterate over the returned generator to receive validated chunks in real time.

Code Reference

Source Location

• Repository: guardrails
• File: guardrails/guard.py
• Lines: L700-721 (StreamRunner creation), L745-795 (__call__ dispatch)

Signature

@trace(name="/guard_call", origin="Guard.__call__")
def __call__(
    self,
    llm_api: Optional[Callable] = None,
    *args,
    prompt_params: Optional[Dict] = None,
    num_reasks: Optional[int] = 1,
    messages: Optional[List[Dict]] = None,
    metadata: Optional[Dict] = None,
    full_schema_reask: Optional[bool] = None,
    **kwargs,  # stream=True passed here
) -> Union[ValidationOutcome[OT], Iterator[ValidationOutcome[OT]]]:

Import

from guardrails import Guard

I/O Contract

Inputs

Name	Type	Required	Description
llm_api	Optional[Callable]	No	LLM callable with streaming support, or model= kwarg
messages	Optional[List[Dict]]	Yes	Chat messages
stream	bool	Yes	Must be True (passed via kwargs)
num_reasks	Optional[int]	No	Should be 0 for streaming (reasks not supported)

Outputs

Name	Type	Description
generator	Iterator[ValidationOutcome[OT]]	Generator yielding ValidationOutcome for each validated chunk

Usage Examples

Streaming Validation

from guardrails import Guard
from guardrails.hub import ToxicLanguage

guard = Guard().use(ToxicLanguage(on_fail="fix"))

stream = guard(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Write a poem about nature."}],
    stream=True,
)

for chunk in stream:
    if chunk.validation_passed:
        print(chunk.validated_output, end="", flush=True)

Related Pages

Implements Principle

• Principle:Guardrails_ai_Guardrails_Stream_Invocation

Uses Heuristic

• Heuristic:Guardrails_ai_Guardrails_Async_Vs_Sync_Validation_Mode