Skip to main content

Prompt Insights

Use this API to retrieve prompt security insights and aggregate metrics for AI agent prompt activity. With this functionality in place, administrators can review prompt events and their policy decisions, filter insights by policy decision, category, severity, and other attributes, and view summary metrics for scanned and redacted prompts over configurable time windows.

Policy decisions indicate whether a prompt was allowed or redacted. Insight categories include anomalies, data uploads, and MCP tool calls. Supported filter fields include policy decision, category, severity, user, agent, reason, rule, and policy.