Smoove Call — agentic threat model
Smoove Call presents a high-risk profile due to its combination of real-time voice capabilities, outbound/inbound telephony access, and 100+ integrations capable of triggering business processes. The lack of explicit security guardrails or compliance certifications in the listing increases the potential for voice-based prompt injection and unauthorized tool execution.
OWASP AIVSS score rationale
| Autonomy of Action | 0.80 | |
| Goal-Driven Planning | 0.70 | |
| Self-Modification | 0.10 | |
| Dynamic Tool Use | 0.80 | |
| Persistent Memory | 0.50 | |
| Contextual Awareness | 0.80 | |
| Dynamic Identity | 0.20 | |
| Multi-Agent Interactions | 0.10 | |
| 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 — likely uses a proprietary or fine-tuned voice-to-voice or LLM + TTS/STT pipeline. Threats include adversarial voice inputs (prompt injection via audio), model reprogramming, or mis-aligned outputs during live calls.
Not certain from the listing — accesses dynamic customer-specific data via live function calling. Threats include data exfiltration of sensitive customer records (PII) and knowledge-base poisoning if RAG is used for support.
Uses live function calling and 100+ integrations to kick off business processes. Threats include insecure tool integration, tool misuse (e.g., unauthorized appointment cancellation or process triggering), and prompt injection leading to unintended function execution.
Not certain from the listing — hosted closed-source solution with low latency (<600ms). Threats include infrastructure compromise, SIP/telephony hacking, and exposure of API keys for the 100+ integrations.
Not certain from the listing — no mention of guardrails or monitoring tools, though it supports transferring to humans. Threats include blind spots in voice-based prompt injection detection and lack of real-time audio transaction logging.
Not certain from the listing — no compliance certifications (like HIPAA, PCI-DSS, SOC2) are mentioned despite handling customer data and business processes. Threats include regulatory non-compliance and unauthorized access to integrated tools.
Not certain from the listing — no explicit multi-agent marketplace interactions mentioned, though it integrates with 100+ external tools. Threats include cascading failures if integrated third-party APIs are compromised.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).