Implementation:Guardrails ai Guardrails ValidationOutcome Structured

Knowledge Sources	Guardrails
Domains	Structured_Output, Data_Conversion
Last Updated	2026-02-14 00:00 GMT

Overview

Concrete pattern for consuming structured validated output from a Guard execution provided by the guardrails package.

Description

When a Guard created via Guard.for_pydantic() executes successfully, the ValidationOutcome.validated_output field contains a Python dict (for single model output) or list (for list model output) that conforms to the Pydantic model's schema. This is the same ValidationOutcome class used for plain text validation, but with OT typed as Dict or List instead of str.

Usage

Access .validated_output on the result of a structured Guard call. Optionally pass it to the Pydantic model constructor for type-safe access.

Code Reference

Source Location

• Repository: guardrails
• File: guardrails/classes/validation_outcome.py
• Lines: L45-51 (validated_output field)

Signature

class ValidationOutcome(IValidationOutcome, ArbitraryModel, Generic[OT]):
    validated_output: Optional[OT] = Field(
        description="The validated, and potentially fixed,"
        " output from the LLM call after passing through validation.",
        default=None,
    )
    # For structured output, OT is Dict or List

Import

from guardrails.classes.validation_outcome import ValidationOutcome

I/O Contract

Inputs

Name	Type	Required	Description
ValidationOutcome	ValidationOutcome[Dict] or ValidationOutcome[List]	Yes	Result from Guard.__call__() with structured output Guard

Outputs

Name	Type	Description
.validated_output	Dict or List	Python dict/list matching Pydantic model schema
.validation_passed	bool	Whether all field validators passed
Pydantic model	BaseModel	Optional: MyModel(**validated_output) for typed access

Usage Examples

Consuming Structured Output

from pydantic import BaseModel, Field
from guardrails import Guard

class Movie(BaseModel):
    title: str = Field(description="Movie title")
    year: int = Field(description="Release year")

guard = Guard.for_pydantic(output_class=Movie)
result = guard(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Name a classic movie."}],
)

if result.validation_passed:
    # Access as dict
    print(result.validated_output["title"])
    print(result.validated_output["year"])

    # Convert to typed Pydantic model
    movie = Movie(**result.validated_output)
    print(movie.title)
    print(movie.year)

Related Pages

Implements Principle

• Principle:Guardrails_ai_Guardrails_Structured_Output_Consumption