Aomni — agentic threat model
Aomni presents a moderate agentic risk profile, primarily driven by its dynamic web-browsing capabilities which expose it to indirect prompt injection and data poisoning. While its lack of direct transactional execution limits physical or financial harm, the ingestion of proprietary product and ICP data makes data confidentiality a key concern.
OWASP AIVSS score rationale
| Autonomy of Action | 0.60 | |
| Goal-Driven Planning | 0.50 | |
| Self-Modification | 0.10 | |
| Dynamic Tool Use | 0.60 | |
| Persistent Memory | 0.40 | |
| Contextual Awareness | 0.70 | |
| Dynamic Identity | 0.10 | |
| Multi-Agent Interactions | 0.20 | |
| Non-Determinism | 0.60 | |
| Opacity & Reflexivity | 0.50 |
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 specific foundation models powering the N-Layer search and chatbot are not disclosed. Threats include model misalignment, prompt injection via untrusted web content during research, and potential model-stealing attacks if proprietary fine-tuning is used.
The agent ingests external web data for research and personalized product/ICP data. This introduces significant risks of data poisoning from malicious web sources (indirect prompt injection) and data exfiltration of sensitive sales/product data via the chatbot or search outputs.
Not certain from the listing — The underlying orchestration framework is not specified. Key threats include insecure tool integration (web browser/search tools) and tool misuse where the agent might be manipulated into executing unintended search queries or actions.
Not certain from the listing — No details are provided regarding hosting, containerization, sandboxing of the web-browsing environment, or API security. Risks include server-side request forgery (SSRF) during web scraping and container breakout if the browsing environment is not isolated.
Not certain from the listing — There is no mention of evaluation frameworks, real-time monitoring, or guardrails to detect hallucinated sales insights or malicious inputs. This creates blind spots regarding drift, prompt injection attempts, and output quality.
Not certain from the listing — Compliance certifications (e.g., SOC 2, GDPR) and specific access controls are not detailed. Given it handles proprietary product data and ICPs, lack of robust tenant isolation and access controls poses a compliance and data privacy risk.
Not certain from the listing — While it offers an API, there is no evidence of a multi-agent ecosystem or third-party agent integrations. The primary risk is unauthorized API access or cascading failures if integrated into broader CRM/sales pipelines.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).