Interface SafetyAssessment

interface SafetyAssessment {
    bias: BiasDetectionResult;
    blendedScore?: number;
    harmfulContent: HarmfulContentGrade;
    hateSpeech: HateSpeechResult;
    llmJudge?: LLMJudgeResponse[];
    method: "llm_judge" | "classifier" | "blended";
    score: number;
    severity: SafetySeverity;
    toxicity: ToxicityResult;
}

Index

Properties

bias blendedScore? harmfulContent hateSpeech llmJudge? method score severity toxicity

Properties

bias

bias: BiasDetectionResult

`Optional`blendedScore

blendedScore?: number

Final blended score incorporating LLM judge

harmfulContent

harmfulContent: HarmfulContentGrade

hateSpeech

hateSpeech: HateSpeechResult

`Optional`llmJudge

llmJudge?: LLMJudgeResponse[]

LLM judge results (if provided)

method

method: "llm_judge" | "classifier" | "blended"

Method used

score

score: number

Overall safety score (0-100, higher = safer)

severity

severity: SafetySeverity

Overall severity

toxicity

toxicity: ToxicityResult

Individual dimension results