Is Your LLM Agent Secure? Scan for Vulnerabilities with Agentic Radar
Large Language Models (LLMs) and agentic workflows are revolutionizing how we interact with technology. But are they secure? Agentic Radar is a security scanner designed to help you find vulnerabilities in your LLM-powered applications. It provides developers, security professionals, and researchers with a clear picture of their agentic systems.
Agentic Radar helps to illuminate potential security risks that might be lurking in your AI systems. Keep reading to find out how this tool can safeguard your projects!
Why You Need a Security Scanner for LLM Agents
Agentic systems are complex, often involving multiple tools and intricate workflows. This complexity makes it challenging to ensure transparency and security. Agentic Radar simplifies this process by providing:
- Workflow Visualization: Graphically represents the agentic system's workflow.
- Tool Identification: Lists all external and custom tools used by the system.
- Vulnerability Mapping: Connects tools to known vulnerabilities, providing an overview of potential risks.
With a comprehensive HTML report, reviewing findings and sharing insights becomes much more manageable. Stay ahead of potential threats!
Key Features of Agentic Radar: Uncover Vulnerabilities
Agentic Radar goes beyond simple scanning. It gives an extensive analysis of your LLM agent's architecture. Here's what makes it special:
- Comprehensive Reporting: Detailed HTML reports with visualizations and vulnerability mappings.
- Security Framework Integration: Maps detected vulnerabilities to OWASP Top 10 LLM Applications and OWASP Agentic AI – Threats and Mitigations frameworks.
- Easy to Use: Simple installation and command-line interface.
By mapping vulnerabilities to established frameworks, Agentic Radar provides context and actionable recommendations.
Getting Started with Agentic Radar: Quick Installation Guide
Ready to start scanning? Here’s how to get Agentic Radar up and running.
Prerequisites: Install Graphviz and Cairo
Agentic Radar relies on Graphviz and Cairo for generating dependency graphs. Install them using the following commands:
-
For macOS (using Homebrew):
-
For Linux (using apt-get):
Installation: Use pip to Install Agentic Radar
Install the package using pip:
With the prerequisites and Agentic Radar installed, you're ready to scan your LLM agents for vulnerabilities.
How to Use Agentic Radar: Scan Your Code in Seconds
Using Agentic Radar is straightforward. The command-line interface makes it easy to integrate into your existing workflows.
Command-Line Options: Customize Your Scan
To see all available options, run:
This command displays a list of options, including:
-i, --input-dir TEXT: Path to the directory containing your code (default: current directory).-o, --output-file TEXT: Specify the output file for the generated report (default:report_YYYYMMDD_HHMMSS.html).
Example Usage: Scanning Your LangGraph Code
To scan code written with LangGraph, use the following command:
This command scans the specified directory and generates a detailed security report in HTML format.
Roadmap: Future Development for Agentic Radar
The Agentic Radar team has exciting plans for the future. Expect the following features in upcoming releases:
- Expanded Framework Support: Including LlamaIndex, Swarm, PydanticAI, AutoGen, and Dify.
- Continuous Integration (CI): Automated code style checks and releases to PyPi.
- Improved Report Design: Enhancements to SVG scaling and overall report aesthetics.
These planned updates will make Agentic Radar an even more powerful and versatile tool for LLM security.
See Agentic Radar in Action: Demo and Community
Want to see Agentic Radar in action? Check out the demo and join the community!
Demo: Google Colab Notebook
Explore the Demo Google Colab Notebook to see how Agentic Radar integrates into your development workflow.
Community: Join the Discord Server
Connect with other developers and security professionals on the Discord community. Discuss features, get support, and contribute to the project.
Contribute to Agentic Radar
We invite contributions from the AI and security community. Improve and innovate Agentic Radar by submitting pull requests, reporting bugs, and suggesting new features.
Agentic Radar: Your First Line of Defense for LLM Security
Don't leave your LLM agentic workflows vulnerable. Use Agentic Radar to scan your code, identify potential risks, and ensure the security of your AI-powered applications. Download it today and take the first step towards a more secure future!
.png)
