guardrails_ai.sdk.guards_api¶

Classes¶

API for managing Guards and running validations.

Module Contents¶

class guardrails_ai.sdk.guards_api.GuardsApi(*, http_client: httpx.AsyncClient, headers: dict[str, str], max_retries: int)¶

Bases: guardrails_ai.sdk.abstract_client.Client

API for managing Guards and running validations.

Accessed via client.guards.

async create(guard: guardrails_ai.types.CreateGuardRequest) → guardrails_ai.types.Guard¶

Creates a Guard on the Guardrails API.

Retries with exponential back-off up to max_retries times on failure.

Parameters:: guard – The request body of the Guard to create.
Returns:: A validated Guard instance.
Raises:: tenacity.RetryError – If all retry attempts are exhausted.

async delete(id: str) → guardrails_ai.types.Guard¶

Deletes a Guard on the Guardrails API.

Retries with exponential back-off up to max_retries times on failure.

Parameters:: id – The unique id of the Guard to delete.
Returns:: A validated Guard instance.
Raises:: tenacity.RetryError – If all retry attempts are exhausted.

async list(*, name: str | None = None) → list[guardrails_ai.types.Guard]¶

List Guards from the Guardrails API.

Retries with exponential back-off up to max_retries times on failure.

Parameters:: name – If provided, will fetch guards only with the designated name. Defaults to None.
Returns:: A validated list of Guard instances.
Raises:: tenacity.RetryError – If all retry attempts are exhausted.

async retrieve(id: str) → guardrails_ai.types.Guard¶

Fetch a Guard by id from the Guardrails API.

Retries with exponential back-off up to max_retries times on failure.

Parameters:: id – The id of the Guard to retrieve.
Returns:: A validated Guard instance.
Raises:: tenacity.RetryError – If all retry attempts are exhausted.

async update(guard: guardrails_ai.types.Guard) → guardrails_ai.types.Guard¶

Updates a Guard on the Guardrails API.

Retries with exponential back-off up to max_retries times on failure.

Parameters:: guard – The updated Guard to persist.
Returns:: A validated Guard instance.
Raises:: tenacity.RetryError – If all retry attempts are exhausted.

async validate(id: str, content: str, **kwargs) → guardrails_ai.types.ValidationOutcome¶

Validate content against a Guard.

Retries with exponential back-off up to max_retries times on failure.

Parameters:

id – The unique id of the Guard to validate against.
content – The text content (typically LLM output) to validate.
**kwargs – Additional fields merged into the request body.

Returns:

A ValidationOutcome describing whether validation passed and, if so, the validated output.

Raises:

tenacity.RetryError – If all retry attempts are exhausted.

chat: guardrails_ai.sdk.chat_completions_api.ChatApi¶

headers: dict[str, str]¶

http_client: httpx.AsyncClient¶

max_retries: int¶