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

    Interface MultilingualAttack

    Multilingual Adversarial Testing — MF-15

    Non-English prompt injection, homoglyph attacks, mixed-language attacks, RTL injection, encoding exploits, and cultural context exploitation.

    Design driver — synthetic scenario, not user research.

    The line below comes from the MiroFish simulation (mirofish-simulation/), whose "practitioners" are model-generated personas. No real person said it and no named individual asked for this module. It is kept because the requirement stands on its own merits, not because anyone requested it.

    "Non-English adversarial testing coverage is obviously insufficient."

    interface MultilingualAttack {
        attackType: string;
        description: string;
        expectedDetection: boolean;
        id: string;
        language: string;
        prompt: string;
    }
    Index
    attackType: string
    description: string
    expectedDetection: boolean
    id: string
    language: string
    prompt: string