Tilores — agentic threat model
Tilores acts as a high-value data aggregator (IdentityRAG) connecting LLMs to sensitive customer databases, presenting a significant data exfiltration and privacy risk if compromised, despite having low autonomous agency.
OWASP AIVSS score rationale
| Autonomy of Action | 0.20 | |
| Goal-Driven Planning | 0.10 | |
| Self-Modification | 0.00 | |
| Dynamic Tool Use | 0.40 | |
| Persistent Memory | 0.30 | |
| Contextual Awareness | 0.60 | |
| Dynamic Identity | 0.10 | |
| Multi-Agent Interactions | 0.00 | |
| Non-Determinism | 0.20 | |
| Opacity & Reflexivity | 0.20 |
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 — Tilores is an IdentityRAG system that connects to external LLMs rather than hosting them. The primary threat at this layer is indirect prompt injection where malicious data retrieved by Tilores manipulates the downstream LLM.
Highly critical layer for Tilores. It centralizes customer data, transactions, and metadata. Key threats include data poisoning of the identity resolution graph, unauthorized data exfiltration of sensitive PII, and lineage/provenance gaps across connected sources.
Not certain from the listing — Tilores acts as a data retrieval tool/API rather than a full agent orchestration framework. Threats include insecure integration of the Tilores API within the consuming LLM's tool-calling framework.
Not certain from the listing — details about hosting, network isolation, and credential storage for connected customer data sources are not provided. Threats include API key exposure and unauthorized access to the Tilores integration endpoints.
Not certain from the listing — there is no mention of built-in logging, monitoring, or guardrails to detect anomalous search queries or unauthorized bulk data extraction attempts.
Not certain from the listing — despite handling highly sensitive customer transaction and order metadata, specific access control policies (RBAC), data minimization practices, or regulatory compliance standards (GDPR/SOC2) are not detailed.
Not certain from the listing — no multi-agent coordination or agent-to-agent marketplace interactions are described for this tool.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).