Agent Wiz is an open-source CLI tool for mapping, analyzing, and threat-modeling AI agent systems. It is installed via pip (repello-agent-wiz) and operates directly from the terminal without requiring additional configuration or vendor lock-in. The tool performs AST-based static analysis of Python orchestrator files to extract real agent-to-tool-to-LLM interaction graphs, making implicit multi-agent workflows visible. These graphs can be exported in .png, .dot, or interactive formats for review and documentation. Threat modeling is conducted using the MAESTRO framework, which evaluates agent interactions across 12 known agentic failure modes. Agent Wiz detects security issues specific to multi-agent architectures, including privilege escalation vectors between agents, prompt confusion, ghost agents, infinite recursion, and trust boundary violations involving memory, tools, and LLMs. Output includes JSON risk summaries, ranked threat paths, and graph-annotated issues structured according to the MAESTRO framework for risk scoring and mitigation guidance. Agent Wiz is framework-agnostic and works out of the box with several major AI agent orchestration frameworks, including Autogen, LangGraph, CrewAI, LlamaIndex, OpenAI SDK, and n8n.