GuardrailDetails Class

Immutable input contract describing a guardrail evaluation.

Constructor

GuardrailDetails(target_type: str, decision_type: str, guardian_name: str | None = None, guardian_id: str | None = None, guardian_provider_name: str | None = None, guardian_version: str | None = None, target_id: str | None = None, decision_reason: str | None = None, decision_code: str | None = None, policy_id: str | None = None, policy_name: str | None = None, policy_version: str | None = None, content_input_hash: str | None = None, content_modified: bool | None = None, external_event_id: str | None = None)

Parameters

Name	Description
target_type Required
decision_type Required
guardian_name	Default value: None
guardian_id	Default value: None
guardian_provider_name	Default value: None
guardian_version	Default value: None
target_id	Default value: None
decision_reason	Default value: None
decision_code	Default value: None
policy_id	Default value: None
policy_name	Default value: None
policy_version	Default value: None
content_input_hash	Default value: None
content_modified	Default value: None
external_event_id	Default value: None

Attributes

target_type

What content is being guarded (e.g., "llm_input", "tool_call").

target_type: str

decision_type

The guardian's decision (e.g., "allow", "deny").

decision_type: str

guardian_name

Human-readable guardian name.

guardian_name: str | None = None

guardian_id

Unique guardian identifier.

guardian_id: str | None = None

guardian_provider_name

Provider name (e.g., "azure.ai.content_safety").

guardian_provider_name: str | None = None

guardian_version

Guardian version string.

guardian_version: str | None = None

target_id

ID of the targeted content.

target_id: str | None = None

decision_reason

Human-readable decision reason.

decision_reason: str | None = None

decision_code

Machine-readable decision code.

decision_code: str | None = None

policy_id

Triggered policy ID.

policy_id: str | None = None

policy_name

Triggered policy name.

policy_name: str | None = None

policy_version

Policy version.

policy_version: str | None = None

content_input_hash

Hash of input content for forensic correlation.

content_input_hash: str | None = None

content_modified

Whether the content was altered by the guardrail.

content_modified: bool | None = None

external_event_id

External event ID for SIEM correlation.

external_event_id: str | None = None