GuardrailFinding Class

A single security finding from a guardrail evaluation.

Constructor

GuardrailFinding(risk_category: str, risk_severity: str, policy_decision_type: str | None = None, policy_id: str | None = None, policy_name: str | None = None, policy_version: str | None = None, risk_score: float | None = None, risk_metadata: list[str] | None = None)

Parameters

Name	Description
risk_category Required
risk_severity Required
policy_decision_type	Default value: None
policy_id	Default value: None
policy_name	Default value: None
policy_version	Default value: None
risk_score	Default value: None
risk_metadata	Default value: None

Attributes

risk_category

Category of risk detected (e.g., "hate_speech", "pii", "jailbreak").

risk_category: str

risk_severity

Severity level (use GuardrailRiskSeverity constants).

risk_severity: str

policy_decision_type

Per-finding decision override.

policy_decision_type: str | None = None

policy_id

Policy that triggered this finding.

policy_id: str | None = None

policy_name

Policy name.

policy_name: str | None = None

policy_version

Policy version.

policy_version: str | None = None

risk_score

Confidence score from 0.0 to 1.0.

risk_score: float | None = None

risk_metadata

Non-PII structural metadata about the finding.

risk_metadata: list[str] | None = None