Product Guides•stable•Updated 2026-05-15

Guardrails — Policies

Configure budget settings, anomaly sensitivity, alert policies, and simulations.

What this page is for

Policies is the configuration surface for workspace policy behavior. It focuses on settings and simulation rather than marketing or general education.

Who uses it

  • Admins
  • Platform operators
  • Security reviewers

What you see

  • Budget settings with daily budget, derived monthly budget, and global status.
  • Anomaly sensitivity set to low, medium, or high.
  • Alert policies with name, type, severity, enabled state, and description.
  • Simulation form and run history.
  • Run detail with draft, summary, result JSON, and baseline vs draft comparison when available.

What you can do

  • Edit the daily budget when permitted.
  • Change anomaly sensitivity.
  • Toggle alert policies.
  • Change severity.
  • Simulate a draft from an existing policy.
  • Open a previous simulation run.

Permissions and safety

  • Budget editing follows budget permissions.
  • Sensitivity and policy changes require policy edit permissions.
  • Simulations should remain visible even when not actionable.

Operational notes

  • The simulation form supports policyId or policyType, enabled, severity, configJson, and baselineWindowDays.
  • The tab should stay focused on admin configuration.

What this page is not for

  • Replacing Alerts routing.
  • Replacing the audit log.
  • Replacing Cost Signals.

Recommended workflow

  1. Confirm that you are in the right workspace and environment.
  2. Use filters before changing configuration or interpreting results.
  3. Open the detail drawer when available.
  4. Check related objects such as alerts, approvals, costs, policies, runs, deliveries, or audit entries.
  5. Apply changes only when the impact is clear.
  6. Re-check the page after the backend refreshes.
  7. Export or copy IDs when you need to escalate.

Related product areas