Paste a shell command or simulate an MCP tool call. See AgentShield's verdict, the rules that fired, the threat taxonomy, and the auditor-defensible compliance mapping — all without installing anything.
Almost 4,000security rules
…community rules to explore
…TP cases
…TN cases
…kingdoms
…compliance frameworks
Live evaluation uses community rules. Browse the published catalog for rule explanations and examples.
AI Agent Lens SaaS adds premium policies, fleet governance, and compliance dashboards.
Explore the demo →
Shell command
AgentShield only evaluates — never executes.
MCP tool call
Arguments
Simulates an MCP JSON-RPC call. No server is contacted.
Browse by threat kingdom
Each kingdom maps to a category in the OWASP / MITRE / ISO threat models. Click an example to evaluate it on the Shell tab.