Guardrails & security
Secret blocking
Stop API keys and secrets from leaking into prompts.
Alongside PII detection, the guardrail layer detects common secret patterns —
such as provider API keys (sk-…) — in prompts and blocks or redacts them
before they reach a model.
- Built-in patterns cover common key formats.
- Add custom org rules for secrets specific to your stack.
- Combine with PII detection for a single pass over each prompt.