As AI agents gain more autonomy and tool access, securing agentic workflows becomes critical. Agentic Radar scans your agent pipeline code and produces a visual map of all tools, data flows, and external connections — making it easy to identify where prompt injection could manipulate behavior, where excessive permissions grant agents unnecessary access, and where sensitive data might leak through tool chains. The scanner understands the specific patterns of popular frameworks.
MCP server detection is a standout feature: Agentic Radar identifies all MCP servers in your configuration, maps their exposed tools and resources, and flags potential security issues like overly broad file system access, unvalidated SQL execution, or authentication gaps. As the MCP ecosystem grows with hundreds of community-built servers, this automated security review becomes essential for any team integrating third-party MCP capabilities.
The tool is open-source and actively maintained by SPLX AI. It runs as a CLI with HTML report output showing interactive diagrams of your agent architecture. Support for CrewAI, LangGraph, AutoGen, and Semantic Kernel covers the major agent frameworks, with more being added. For teams building production agent systems, Agentic Radar provides the security audit layer that is otherwise completely manual and error-prone.