Rootly — agentic threat model
Rootly acts as a high-privilege incident management and deployment safety agent integrated with Claude Code. Its access to production deployment checks and incident response workflows presents a high-impact risk profile if compromised via prompt injection or tool abuse.
OWASP AIVSS score rationale
| Autonomy of Action | 0.70 | |
| Goal-Driven Planning | 0.80 | |
| Self-Modification | 0.20 | |
| Dynamic Tool Use | 0.80 | |
| Persistent Memory | 0.60 | |
| Contextual Awareness | 0.80 | |
| Dynamic Identity | 0.50 | |
| Multi-Agent Interactions | 0.60 | |
| 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 — relies on Claude Code (Anthropic Claude models) via MCP. Threat of prompt injection leading to unauthorized incident creation, false alerts, or deployment safety bypass.
Not certain from the listing — accesses incident logs, retrospectives, and on-call schedules. Risk of data exfiltration of sensitive system architecture details or PII revealed during active incidents.
Uses MCP/API-backed skills and commands driving the Rootly platform. Risks include insecure tool integration where Claude Code is tricked into executing destructive API calls or modifying on-call schedules.
Not certain from the listing — runs as an MCP server/plugin. Risks include exposure of Rootly API keys or local execution environment compromise if the host running Claude Code is exploited.
Not certain from the listing — no explicit guardrails or evaluation metrics mentioned. Risk of undetected drift in automated deploy-safety checks leading to false negatives.
Not certain from the listing — relies on Rootly's platform auth and Claude Code's local permissions. Risk of privilege escalation if the MCP token has over-privileged access to production systems.
Integrates directly with Claude Code (multi-agent/ecosystem interaction). Risk of cascading failures where a compromised Claude Code agent abuses Rootly commands to disrupt operations or trigger false incidents.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).