Chyo — agentic threat model
Chyo is a customer-facing AI agent platform with moderate risk, primarily driven by its 'automatic learning' capabilities from multiple sources which expose it to data poisoning, and its multi-agent creation features which increase the internal attack surface.
OWASP AIVSS score rationale
| Autonomy of Action | 0.50 | |
| Goal-Driven Planning | 0.30 | |
| Self-Modification | 0.20 | |
| Dynamic Tool Use | 0.40 | |
| Persistent Memory | 0.40 | |
| Contextual Awareness | 0.50 | |
| Dynamic Identity | 0.10 | |
| Multi-Agent Interactions | 0.60 | |
| Non-Determinism | 0.60 | |
| Opacity & Reflexivity | 0.70 |
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 — The underlying foundation models are not specified. As a closed-source customer service tool, it likely relies on commercial LLMs, making it susceptible to prompt injection, model alignment bypasses, and indirect prompt injection via customer inputs.
The platform features 'Automatic Learning from multiple sources' and 'Fine tuning of FAQ'. This introduces a high risk of data and knowledge-base poisoning if malicious or untrusted external sources are ingested, potentially leading to the generation of toxic, inaccurate, or malicious customer responses.
Not certain from the listing — The specific orchestration framework is proprietary. However, the system manages state for product recommendations, multi-currency, and human handover, indicating potential risks around insecure tool integration and state manipulation during active sessions.
Not certain from the listing — No deployment, hosting, or sandboxing details are provided. As a closed-source SaaS, standard web application vulnerabilities and tenant isolation risks apply.
Not certain from the listing — While 'Handover to human' is supported as a fallback mechanism, the listing does not detail active guardrails, real-time anomaly detection, or automated evaluation metrics to prevent drift or offensive outputs.
Not certain from the listing — No compliance certifications (e.g., SOC2, ISO 27001) or specific data privacy controls (e.g., GDPR/CCPA masking for customer chats) are mentioned in the public directory listing.
The platform allows users to 'Create multiple agents'. This introduces multi-agent ecosystem risks, such as cross-agent trust abuse, cascading failures if one agent is compromised, and authorization boundary confusion between different configured agents.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).