Agent Maturity Compass API - v1.2.0
    Preparing search index...

    Interface ParametricFactualityInput

    Factuality Dimensions — inspired by Google DeepMind FACTS benchmark

    Three orthogonal factuality axes:

    1. Parametric factuality — accuracy of the model's internal knowledge
    2. Search/retrieval factuality — accuracy when using RAG or search tools
    3. Grounded factuality — accuracy when answering from provided context

    Each dimension is scored 0-100 with evidence-based assessment.

    interface ParametricFactualityInput {
        expressesUncertainty: boolean;
        refusesWhenUnsure: boolean;
        totalClaims: number;
        unverifiable: number;
        verifiedCorrect: number;
        verifiedIncorrect: number;
    }
    Index
    expressesUncertainty: boolean

    Whether the agent expresses calibrated uncertainty

    refusesWhenUnsure: boolean

    Whether the agent refuses to answer when unsure

    totalClaims: number

    Total knowledge claims made from internal knowledge

    unverifiable: number

    Claims that could not be verified

    verifiedCorrect: number

    Claims verified as correct

    verifiedIncorrect: number

    Claims verified as incorrect (hallucinations)