Agent-Wiz

One-Minute Overview#

Agent Wiz is a Python CLI tool for extracting agentic workflows from popular AI frameworks and performing automated threat assessments using established threat modeling methodologies. Built for developers, researchers, and security teams, Agent Wiz brings visibility to complex LLM-based orchestration to visualize flows, map tool/agent interactions, and generate actionable security reports.

Core Value: Transforms complex and often invisible AI agent workflows into visual security analysis, enabling teams to proactively identify and mitigate system risks.

Getting Started#

Installation Difficulty: Low - Simple pip installation with only OpenAI API key configuration required

# Installation command
pip install repello-agent-wiz

Configure your OpenAI API key:

# Option 1: Environment variable
export OPENAI_API_KEY=sk-...

# Option 2: .env file (recommended)

Is this suitable for me?

• ✅ AI Agent System Security Auditing: When you need to assess the security of LLM-based agent systems
• ✅ Complex Workflow Visualization: When you need to understand and document interactions between multiple agents
• ✅ Security Compliance Checking: When your organization needs to meet AI security compliance requirements
• ❌ Simple AI Application Development: If you're just building simple AI applications without deep security analysis needs

Core Capabilities#

1. Workflow Extraction - Understanding Agent Architecture from Code#

Uses AST-based static parsing to extract agent workflows from code without modifying existing code. Real Value: Transforms complex agent code structures into visual relationship graphs, helping teams quickly understand system architecture.

2. Threat Vector Visualization - Showing System Interaction Patterns#

Interactive graph displays connections and call chains between agents, agents and tools. Real Value: Visually demonstrates system attack surfaces and data flow paths, helping to identify potential security risk points.

3. Automated Threat Assessment - Generating Security Reports#

Generates comprehensive security assessment reports using AI agent threat modeling frameworks like MAESTRO. Real Value: Provides standardized security analysis to help teams systematically identify and address vulnerabilities.

4. Framework Agnostic Design - Supporting Mainstream AI Frameworks#

Supports all major LLM orchestration frameworks including Autogen, AgentChat, CrewAI, LangGraph, etc. Real Value: Single tool can analyze systems built with multiple frameworks, eliminating the need to learn multiple analysis methods.

5. Developer Friendly - Simple and Easy to Use#

Offers simple CLI interface, extensible SDK, and clean JSON export capabilities. Real Value: Reduces the technical barrier to security analysis, allowing non-security experts to participate in AI system security assessment.

Technology Stack & Integration#

Development Language: Python Main Dependencies: OpenAI API Integration Method: CLI Tool / SDK

Ecosystem & Extensions#

• Framework Support: Currently supports mainstream AI frameworks like Autogen, AgentChat, CrewAI, LangGraph, LlamaIndex, n8n, OpenAI Agents, Pydantic-AI, Swarm, Google-ADK
• Extensibility: AST-based static parsers can be easily extended to support more frameworks

Maintenance Status#

• Development Activity: Actively maintained with multi-framework parsing support and ongoing security model extensions
• Recent Updates: Recent development with more threat model analysis (STRIDE, PASTA, LINDDUN) under development
• Community Response: Welcomes contributions with detailed contribution guidelines provided

Commercial & Licensing#

License: Apache 2.0

• ✅ Commercial Use: Allowed
• ✅ Modification: Allowed
• ⚠️ Restrictions: Must include copyright notice

Documentation & Learning Resources#

• Documentation Quality: Comprehensive
• Official Documentation: GitHub repository README
• Example Code: Available (examples/code directory)
• Learning Curve: Low to medium, requires basic Python knowledge and understanding of AI agent frameworks