Playground and Monitor

Playground

Dry-run built-in steps against a sample prompt without spending provider tokens.

• Choose a policy, or “All configured guardrails”
• Paste a prompt (and optional completion for POST / BOTH cards)
• Inspect which steps passed, flagged or blocked

Custom and vendor steps are listed as informational — they are not executed in the dry-run engine.

Monitor

Live guardrail metrics for the selected time window (same range picker as Usage):

• Evaluations, blocked requests, pass rate
• Average latency added
• Active guardrails
• Daily chart
• CSV export (/usage/export?…&tab=guardrails)

Pair Monitor with Request Logs filtered by outcome=blocked (or a specific guardrailId / policyId) when you need the raw prompt.