Botpress — agentic threat model
Botpress is a highly capable, enterprise-grade agent platform with deep integrations into business systems, presenting a significant attack surface if prompt injection or insecure tool execution allows unauthorized access to connected databases and helpdesks.
OWASP AIVSS score rationale
| Autonomy of Action | 0.70 | |
| Goal-Driven Planning | 0.60 | |
| Self-Modification | 0.20 | |
| Dynamic Tool Use | 0.70 | |
| Persistent Memory | 0.60 | |
| Contextual Awareness | 0.70 | |
| Dynamic Identity | 0.30 | |
| Multi-Agent Interactions | 0.40 | |
| 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 — Botpress is model-agnostic and supports various LLMs. Threats include adversarial prompt injection bypassing support guardrails, or model reprogramming to output malicious content to customers.
Not certain from the listing — Integrates with existing business systems and data sources for RAG. Threats include knowledge-base poisoning via public-facing support channels or unauthorized data exfiltration of sensitive customer PII.
Botpress provides the core orchestration framework with developer-level control for multi-turn conversations. Threats include insecure tool integration with business systems, prompt injection leading to unauthorized tool execution, and state manipulation.
Not certain from the listing — Deploys across multiple communication channels and can be hosted or standalone. Threats include container compromise, exposed webhook endpoints, and insecure credential storage for third-party integrations.
Not certain from the listing — The listing does not detail specific evaluation or guardrail features, though it is an enterprise platform. Gaps could lead to undetected model drift or prompt injection attacks going unnoticed.
Not certain from the listing — Mentions 'enterprise-grade' but lacks specific details on RBAC, compliance certifications (like SOC2), or audit logging in the directory listing.
Botpress is a massive platform that has shipped over 750,000 agents. Threats include supply-chain risks from community integrations/templates, and cascading failures if multiple deployed agents interact or share resources.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).