Cyiro
Sign InRequest access

Docs center

Self-hosted docs for Guardrails for AI.

Technical guidance for observability, guardrails, permissioning, and automation in one first-party documentation surface.

Self-hosted docs

Policy Authoring

Create, test, and deploy guardrail policies that enforce safe and compliant AI operations.

docs.cyiro.comproduction-ready guidance

Policy structure

Define policies with clear conditions, thresholds, and enforcement actions.

  • Example: Policy components: condition, threshold, enforcement_action, notification_targets
  • Example: Runtime policy: unsafe_output_detection with confidence_threshold: 0.85
  • Example: Change policy: pricing_increase_guardrail with percentage_threshold: 10%

Testing and validation

Test policies against historical data and edge cases before deploying to production.

  • Example: Test policies using historical runtime events and change records
  • Example: Validate policy logic with unit tests and integration scenarios
  • Example: Deploy to staging first, then production with monitoring