Policy & compliance

Model Governance

Govern which models an agent may call — not just which APIs. Handle export controls, frontier model restrictions, and emergency suspensions without redeploying.

When you need this

ScenarioIRIS control
Standard models onlyDefault registry
Frontier / cyber-capable modelsTier + export-control + HITL gates
Government or provider suspends a modelDirective kill switch + auto-fallback
Audit trail requiredEvidence Vault

Cedar model rules

IRIS enforces model governance via Cedar rules evaluated at every inference call:

RuleSeverityTrigger
IRIS-MODEL-001CRITICALActive suspension directive (no fallback)
IRIS-MODEL-002HIGHModel not in passport allowed_models
IRIS-MODEL-003HIGHModel tier not in allowed_model_tiers
IRIS-MODEL-004CRITICALExport-control: invalid work authorization
IRIS-MODEL-005HIGHFrontier model requires HITL in staging/production

Passport allowlists

Set allowed_models and allowed_model_tiers in passport.yaml to restrict which models an agent may call. The registry provides tier metadata; the passport is the per-agent allowlist.

Model registry

Edit governance/models/registry.yaml to declare tiers, export-control status, HITL requirements, and fallback chains.

iris models list
iris models reload

Directive kill switches

When a model must be suspended org-wide (e.g. government export-control directive), edit governance/directives/active.yaml and merge via PR:

- directive_id: bis-2026-0612-fable
  model_id: claude-fable-5
  status: suspended
  fallback_model: claude-sonnet-4-6

IRIS hot-reloads on the next inference call. No application redeploy.

iris models directives

IrisAnthropic with auto-fallback

from iris_anthropic import IrisAnthropic

client = IrisAnthropic(
  passport=passport,
  user_work_authorization="us-citizen",
  auto_fallback=True,
)

Full guide on GitHub

For the complete reference including passport allowlists and tier definitions:

MODEL_GOVERNANCE.md on GitHub ↗