assistant-ui — agentic threat model
assistant-ui is a frontend React library rather than an autonomous agent, meaning its direct agentic risk is negligible; however, client-side security risks like XSS or data exposure exist if it is integrated insecurely with backend LLM orchestrators.
OWASP AIVSS score rationale
| Autonomy of Action | 0.00 | |
| Goal-Driven Planning | 0.00 | |
| Self-Modification | 0.00 | |
| Dynamic Tool Use | 0.00 | |
| Persistent Memory | 0.00 | |
| Contextual Awareness | 0.10 | |
| Dynamic Identity | 0.00 | |
| Multi-Agent Interactions | 0.00 | |
| Non-Determinism | 0.10 | |
| Opacity & Reflexivity | 0.10 |
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 — assistant-ui is a frontend library and does not bundle or run foundation models directly, though it connects to them via SDKs.
Not certain from the listing — The library handles UI rendering of chat messages and does not manage vector databases or RAG data operations directly.
Not certain from the listing — While it integrates with Langchain and Vercel AI SDK, the library itself is a UI layer and does not orchestrate agent planning or tool execution.
Not certain from the listing — As a React library, deployment depends entirely on the host application's frontend infrastructure and has no built-in hosting or sandboxing.
Not certain from the listing — The library focuses on UI rendering and does not provide LLM evaluation, guardrails, or observability logging out of the box.
Not certain from the listing — Security controls like authentication, authorization, and input sanitization (beyond basic markdown rendering) must be implemented by the parent application.
Not certain from the listing — The library does not participate in multi-agent marketplaces or ecosystem-level orchestrations directly, acting only as the presentation layer.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).