Vivi — agentic threat model
Vivi's primary risk lies in its handling of highly sensitive mental health data over a public messaging channel (WhatsApp) and the potential for LLM-based conversational drift to provide harmful advice to vulnerable users.
OWASP AIVSS score rationale
| Autonomy of Action | 0.40 | |
| Goal-Driven Planning | 0.50 | |
| Self-Modification | 0.20 | |
| Dynamic Tool Use | 0.20 | |
| Persistent Memory | 0.70 | |
| Contextual Awareness | 0.80 | |
| Dynamic Identity | 0.00 | |
| Multi-Agent Interactions | 0.00 | |
| Non-Determinism | 0.70 | |
| Opacity & Reflexivity | 0.60 |
Scored with the canonical OWASP AIVSS formula (AIVSS calculator reference); agentic risk factors estimated from the agent’s described capabilities.
MAESTRO 7-layer threat model
Per-layer threats for this agent. Layers tagged “not certain from listing” are general, caveated commentary where the public description didn’t pin that layer.
Not certain from the listing — Vivi likely uses a fine-tuned or heavily prompted commercial or open-source LLM optimized for empathy and clinical protocols. Threats include prompt injection bypassing clinical safety guardrails or generating harmful/triggering advice.
Not certain from the listing — Vivi processes highly sensitive personal health information (PHI) and DASS-21 scores. Threats include unauthorized access to vector databases storing user conversation history or data exfiltration via WhatsApp.
Not certain from the listing — The orchestration framework manages state across WhatsApp sessions and triggers clinical screening flows. Threats include state manipulation or session hijacking, leading to incorrect clinical assessments.
Not certain from the listing — Vivi is deployed on WhatsApp, requiring integration with the WhatsApp Business API. Threats include insecure webhook endpoints, API key exposure, or compromise of the hosting environment processing PHI.
Vivi utilizes continuous clinical oversight to monitor and evolve its conversational safety. However, real-time automated guardrails are critical to prevent toxic or harmful outputs during 24/7 unmonitored WhatsApp interactions.
Not certain from the listing — While Vivi claims to ensure data privacy, handling mental health data requires strict compliance with regulations like HIPAA or GDPR. The listing does not specify formal compliance certifications.
Vivi operates as a standalone vertical agent on WhatsApp and does not appear to interact with an external multi-agent ecosystem, minimizing cascading agent-to-agent trust threats.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).
These scores are auto-generated from public information (the agent's own listing, docs, and repository) using the canonical OWASP AIVSS formula and the MAESTRO framework — an estimate for guidance, not a penetration test, audit, or certification. See the scoring methodology. Are you the vendor? Factual corrections are free.