About Agentic Radar
Agentic Radar is an open source security scanner from SPLX for LLM agent workflows. Pointed at a project folder, it analyzes the agent system and produces an HTML security report that covers how the workflow is built, which tools it uses and where known vulnerabilities apply.
It is built for developers, researchers and security professionals who need to understand and assess agentic systems before they ship. The report includes a workflow visualization graph, a list of every external and custom tool the system uses, detection of the MCP servers each agent relies on, and a vulnerability map that links identified tools to known weaknesses. Findings are mapped to the OWASP Top 10 for LLM Applications and the OWASP Agentic AI threats and mitigations guidance. The scan command supports LangGraph, CrewAI, n8n, OpenAI Agents and AutoGen projects. A test command runs live vulnerability tests against OpenAI Agents workflows, and an optional prompt hardening feature rewrites detected system prompts into structured prompts for OpenAI Agents, CrewAI and AutoGen. It installs with pip and can run inside a CI/CD workflow.
The scanner is free and open source on GitHub. No paid tier or hosted version of the scanner is published.
A companion Agentic Visualizer is available for teams that only want the workflow graph without the security analysis.