The SplxAI Team is excited to announce the release of Agentic Radar, our open-source contribution to the AI security community. As AI systems grow more autonomous, the need for transparency, security, and explainability becomes more urgent. With Agentic Radar, we are taking a significant step toward securing agentic AI workflows by providing practitioners with a powerful tool to gain deep insights into decision making paths of AI systems and their security vulnerabilities. This will help security teams meet requirements of AI compliance policies, which demand explainability of AI systems and disclosure of the AI-BOM (AI Bill of Materials) within them.
What is Agentic Radar?
Agentic Radar is an open-source scanner tool for agentic systems that helps security teams and AI engineers understand how AI agents interact with tools, external components, and with each other. By visualizing an AI system’s architecture through static code analysis, it reveals hidden workflows and potential vulnerabilities, allowing security teams to secure them proactively. The tool supports a variety of agentic frameworks and our team will be constantly shipping more integrations.
Agentic Radar enables AI security practitioners to:
Visualize AI workflows: Generate a graph of an AI system’s components – showing how agents and tools form decision paths.
Identify external tools: Detect all tools, APIs, and services integrated within the workflow.
Map AI vulnerabilities: Identify potential vulnerabilities in agentic workflows and align the findings with these LLM security frameworks:
See instant remediation steps: Get clear and actionable fixes to mitigate risks and strengthen the security of your agentic systems.
The results of Agentic Radar's workflow assessments are delivered in an HTML report for easy access and distribution. Below you can see an example of a visualized agentic workflow graph:
Tool vulnerabilities are shown with a detailed description, security framework mappings, and remediation steps for instant response:
Let's stop the guesswork in Agentic AI Security
Securing agentic AI workflows starts with transparency. Without understanding how agents, tools, and data flows interact, it's impossible to conduct precise security testing or ensure compliance. Agentic Radar is the first tool of its kind, giving AI security teams real visibility into agentic workflows and potential vulnerabilities, enabling targeted risk assessments and a more robust AI security posture.
At SplxAI, we believe securing complex AI systems should be accessible and efficient. We’re committed to supporting the AI security community, which is why we decided to make Agentic Radar fully open source.
Try it out for yourself – scan the source code of your own agentic system and see the results firsthand. Our public repo includes a detailed guide and some great examples to get started. Feel free to leave a star if you want to support our cause for AI transparency and secure agentic workflows!
Table of contents
