Aikido Security — agentic threat model
The Aikido Security MCP plugin introduces significant risk by granting an LLM agent (Claude Code) direct access to codebase vulnerabilities and secrets, creating a high-impact vector for code exfiltration or malicious code injection if the agent or MCP channel is compromised.
OWASP AIVSS score rationale
| Autonomy of Action | 0.40 | |
| Goal-Driven Planning | 0.50 | |
| Self-Modification | 0.10 | |
| Dynamic Tool Use | 0.60 | |
| Persistent Memory | 0.20 | |
| Contextual Awareness | 0.70 | |
| Dynamic Identity | 0.30 | |
| Multi-Agent Interactions | 0.60 | |
| Non-Determinism | 0.40 | |
| 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 — The specific foundation models powering Claude Code or the underlying analysis are not detailed in this plugin's directory listing.
Not certain from the listing — While the plugin reads the working codebase and scans for secrets/vulnerabilities, the exact data storage, vector database, or RAG pipeline details of the MCP server are not specified.
The agent uses the Model Context Protocol (MCP) to integrate with Claude Code, orchestrating tool calls to trigger scans and read findings. Threats include tool misuse (e.g., unauthorized scan triggers or reading sensitive findings) and insecure tool integration.
Not certain from the listing — The MCP server runs locally or remotely to interface with Aikido's cloud, but the exact sandboxing, containerization, or network isolation of the MCP server is not detailed.
Not certain from the listing — The listing does not detail the logging, guardrails, or observability mechanisms of the MCP integration itself, though it reads AppSec findings.
Not certain from the listing — The plugin handles highly sensitive data (secrets detection, SAST findings, codebase access). However, specific authentication, authorization, or compliance controls (like SOC2 or RBAC for the MCP server) are not detailed in the listing.
The plugin operates in a multi-agent/ecosystem context by connecting Claude Code (the primary agent) to the Aikido MCP server (the secondary tool/agent). Threats include A2A trust abuse, where a compromised Claude Code session could abuse the Aikido MCP server to exfiltrate codebase vulnerabilities or secrets.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).