Guard

The Guard class.

This class is the main entry point for using Guardrails. It is initialized from one of the following class methods:

• from_rail
• from_rail_string
• from_pydantic
• from_string

The __call__ method functions as a wrapper around LLM APIs. It takes in an LLM API, and optional prompt parameters, and returns the raw output from the LLM and the validated output.

Initialize the Guard.

state property

state: GuardState

Return the state.

from_rail classmethod

from_rail(rail_file: str, num_reasks: Optional[int] = None) -> Guard

Create a Schema from a .rail file.

Parameters:

Name	Type	Description	Default
rail_file	str	The path to the .rail file.	required
num_reasks	Optional[int]	The max times to re-ask the LLM for invalid output.	None

Returns:

Type	Description
Guard	An instance of the Guard class.

from_rail_string classmethod

from_rail_string(rail_string: str, num_reasks: Optional[int] = None) -> Guard

Create a Schema from a .rail string.

Parameters:

Name	Type	Description	Default
rail_string	str	The .rail string.	required
num_reasks	Optional[int]	The max times to re-ask the LLM for invalid output.	None

Returns:

Type	Description
Guard	An instance of the Guard class.

from_pydantic classmethod

from_pydantic(output_class: Type[BaseModel], prompt: Optional[str] = None, instructions: Optional[str] = None, num_reasks: Optional[int] = None) -> Guard

Create a Guard instance from a Pydantic model and prompt.

from_string classmethod

from_string(validators: List[Validator], description: Optional[str] = None, prompt: Optional[str] = None, instructions: Optional[str] = None, reask_prompt: Optional[str] = None, reask_instructions: Optional[str] = None, num_reasks: Optional[int] = None) -> Guard

Create a Guard instance for a string response with prompt, instructions, and validations.

Arguments

Name	Type	Description	Default
validators	List[Validator]	(List[Validator]): The list of validators to apply to the string output.	required
description	str	A description for the string to be generated. Defaults to None.	None
prompt	str	The prompt used to generate the string. Defaults to None.	None
instructions	str	Instructions for chat models. Defaults to None.	None
reask_prompt	str	An alternative prompt to use during reasks. Defaults to None.	None
reask_instructions	str	Alternative instructions to use during reasks. Defaults to None.	None
num_reasks	int	The max times to re-ask the LLM for invalid output.	None

configure

configure(num_reasks: Optional[int] = None)

Configure the Guard.

__call__

__call__(llm_api: Union[Callable, Callable[[Any], Awaitable[Any]]], prompt_params: Optional[Dict] = None, num_reasks: Optional[int] = None, prompt: Optional[str] = None, instructions: Optional[str] = None, msg_history: Optional[List[Dict]] = None, metadata: Optional[Dict] = None, full_schema_reask: Optional[bool] = None, *args, **kwargs) -> Union[Tuple[Optional[str], Any], Awaitable[Tuple[Optional[str], Any]]]

Call the LLM and validate the output. Pass an async LLM API to return a coroutine.

Parameters:

Name	Type	Description	Default
llm_api	Union[Callable, Callable[[Any], Awaitable[Any]]]	The LLM API to call (e.g. openai.Completion.create or openai.Completion.acreate)	required
prompt_params	Optional[Dict]	The parameters to pass to the prompt.format() method.	None
num_reasks	Optional[int]	The max times to re-ask the LLM for invalid output.	None
prompt	Optional[str]	The prompt to use for the LLM.	None
instructions	Optional[str]	Instructions for chat models.	None
msg_history	Optional[List[Dict]]	The message history to pass to the LLM.	None
metadata	Optional[Dict]	Metadata to pass to the validators.	None
full_schema_reask	Optional[bool]	When reasking, whether to regenerate the full schema or just the incorrect values. Defaults to True if a base model is provided, False otherwise.	None

Returns:

Type	Description
Union[Tuple[Optional[str], Any], Awaitable[Tuple[Optional[str], Any]]]	The raw text output from the LLM and the validated output.

parse

parse(llm_output: str, metadata: Optional[Dict] = None, llm_api: Optional[Callable] = None, num_reasks: Optional[int] = None, prompt_params: Optional[Dict] = None, full_schema_reask: Optional[bool] = None, *args, **kwargs) -> Union[Any, Awaitable[Any]]

Alternate flow to using Guard where the llm_output is known.

Parameters:

Name	Type	Description	Default
llm_output	str	The output being parsed and validated.	required
metadata	Optional[Dict]	Metadata to pass to the validators.	None
llm_api	Optional[Callable]	The LLM API to call (e.g. openai.Completion.create or openai.Completion.acreate)	None
num_reasks	Optional[int]	The max times to re-ask the LLM for invalid output.	None
prompt_params	Optional[Dict]	The parameters to pass to the prompt.format() method.	None
full_schema_reask	Optional[bool]	When reasking, whether to regenerate the full schema or just the incorrect values.	None

Returns:

Type	Description
Union[Any, Awaitable[Any]]	The validated response. This is either a string or a dictionary, determined by the object schema defined in the RAILspec.