Twig — agentic threat model
Twig presents a moderate risk profile primarily centered on data privacy and integrity, as it ingests sensitive internal tickets and documentation to autonomously answer user queries, making it susceptible to RAG-based data exfiltration and prompt injection.
OWASP AIVSS score rationale
| Autonomy of Action | 0.40 | |
| Goal-Driven Planning | 0.20 | |
| Self-Modification | 0.00 | |
| Dynamic Tool Use | 0.20 | |
| Persistent Memory | 0.20 | |
| Contextual Awareness | 0.50 | |
| Dynamic Identity | 0.10 | |
| Multi-Agent Interactions | 0.10 | |
| 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.
Twig utilizes generative AI and LLMs to power support bots, exposing them to standard foundation model threats such as adversarial prompt injection and misaligned outputs.
The agent ingests documentation, tutorials, and customer tickets, creating a significant risk of data poisoning from malicious tickets and data exfiltration of sensitive customer PII.
Not certain from the listing — the specific orchestration framework is not disclosed, but the agent orchestrates RAG pipelines over tickets and docs, risking prompt injection and insecure retrieval.
Not certain from the listing — hosting infrastructure, sandboxing of the Agent Factory, and secrets management for ticket integrations are not detailed.
Not certain from the listing — no explicit mention of guardrails, evaluation metrics, or observability tools for monitoring drift or toxic outputs.
Not certain from the listing — compliance certifications (like SOC2) or access control mechanisms for sensitive ticket data are not specified.
Not certain from the listing — there is no mention of multi-agent coordination or external agent-to-agent ecosystems beyond the 'Agent Factory' creation process.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).