Skip to main content

Guardrail Request Lifecycle

Guardrails can evaluate content at multiple stages before and after model processing.

Core Building Blocks

Providers

Providers supply detection capabilities. Bud AI Foundry supports Bud Sentinel and cloud moderation providers.

Probes

A probe targets a specific risk area (for example, PII, toxicity, jailbreak, or safety categories).

Rules

Rules are the detection units within probes. You can often enable/disable or tune behavior per rule.

Profiles

A profile is your reusable policy package:
  • selected probes
  • guard type scope
  • severity threshold and overrides
  • deployment targets

Governance Policies

A governance policy is an agentic workflow policy package. It contains rules that combine checks, lifecycle hooks, outcomes, optional context filters, guidance, and priority. Governance policies are deployed to projects and attached to agent versions.

Deployments

Deployment activates the profile or governance policy in runtime paths, either as standalone, attached to an existing endpoint, or made available to agent versions in a project.

Guard Types and Evaluation Scope

Severity and Policy Tuning

Higher thresholds generally increase sensitivity and can reduce unsafe leakage, but may increase false positives. Calibrate with real prompts and staged rollout.

Execution Patterns

Operational Views in Bud Admin

  • Guardrails page: profile discovery, filtering, and creation entry point.
  • Guardrail details: General, Probes, and Deployments tabs.
  • Project Guardrails tab: project-scoped list with status and management actions.

Governance Policy Rule Anatomy

Governance policies are especially useful when multiple models, tools, data sources, and actions participate in a workflow and each organization needs its own security or compliance rules.