Scan agent workflows for tools, MCP exposure, and adversarial risk with Agentic Radar
Use Agentic Radar to statically scan agent workflows, map tools and MCP servers, generate shareable security reports, and optionally run adversarial runtime tests before rollout.
npx skills add agentskillexchange/skills --skill scan-agent-workflows-for-tools-mcp-exposure-and-adversarial-risk-with-agentic-radar
pip install agentic-radar. For framework-specific extras, use pip install "agentic-radar[crewai]" or pip install "agentic-radar[openai-agents]" when needed. Run agentic-radar scan <framework> -i <path> -o report.html to generate a report, or agentic-radar test openai-agents "<entrypoint>" for runtime adversarial testing.Use Agentic Radar when the job is to inspect an agent workflow before release, inventory its tools and MCP servers, map likely vulnerabilities, and produce a reviewable report for security or engineering follow-up. The upstream project supports explicit framework-targeted scan commands such as `agentic-radar scan langgraph`, `scan crewai`, `scan n8n`, `scan openai-agents`, and `scan autogen`, plus a runtime `test` mode for adversarial checks in supported workflows.
What this skill actually does
Invoke this instead of using the product normally when you need a repeatable pre-deployment review step, not just a generic security platform. The operator workflow is concrete: point Agentic Radar at a workflow, choose the framework, generate the HTML report, inspect MCP and tool findings, and optionally run the built-in vulnerability test suite against an entrypoint.
The scope boundary that keeps this skill-shaped is narrow and operational: workflow security inspection and targeted agent testing. It is not a generic agent framework listing, not a plain product card, and not a broad SDK entry. The user is invoking a specific audit workflow with clear inputs, outputs, and stop conditions.