governance-zones

Governance Zones

Governance zones define the operational environment for an agent. They allow different governance strictness for development, staging, and production contexts without maintaining separate configs.

Built-In Zones

Zone	Description
development	Relaxed governance for local development and testing
staging	Moderate governance mirroring production behavior
production	Full governance enforcement
sandbox	Isolated environment with maximum restrictions

Configuring a Zone

agents:
  my-agent:
    governance_zone: production

Zone Overrides

governance_zones:
  development:
    allow_threshold: 0.5
    audit_level: minimal
  production:
    allow_threshold: 0.75
    audit_level: full
    require_birth_certificate: true

[!warning]

Agents operating outside their declared governance zone can be flagged by the Isolation Integrity dimension. Always ensure governance_zone in the config matches the actual deployment environment.

Zone in Birth Certificates

The governance zone is recorded in the agent's Birth Certificate at creation time and becomes part of its cryptographic identity.

Related

• Agent Birth Certificates

• Configuration Overview

• Organization Governance

---

Page: Verdicts

Verdicts

Every governance evaluation produces one of five verdicts.

Verdict	Meaning
ALLOW	Action is permitted. Proceed.
DENY	Action is not permitted. Do not proceed.
MODIFY	Action is permitted with modifications. Proceed with reduced scope or additional constraints.
ESCALATE	Action requires human review before proceeding.
SUSPEND	Agent is suspended. No further actions permitted until reviewed.

ALLOW

The action passed governance. UCS exceeded the allow threshold and no vetoes fired.

if verdict.verdict == Verdict.ALLOW:
    result = tool.execute(action)

DENY

The action failed governance. Either a veto dimension fired (Tier 1) or UCS fell below the deny threshold (Tier 2).

if verdict.verdict == Verdict.DENY:
    print(f"Denied: {verdict.reasoning}")
    print(f"Vetoed by: {verdict.vetoed_by}")

MODIFY

The action is permitted but with constraints. The verdict includes a modifications dict describing what must change.

if verdict.verdict == Verdict.MODIFY:
    modifications = verdict.modifications
    # e.g. {"reduce_scope": True, "require_confirmation": True}

ESCALATE

The action requires human review. Returned when the Human Override dimension vetoes, when trust is too low for Tier 3 to decide, or when configured escalation rules trigger.

SUSPEND

The agent has been suspended. Returned when multiple serious violations occur in sequence or when configured suspension thresholds are exceeded.

[!warning]

A SUSPEND verdict means the agent cannot take any further actions until a human reviews and reinstates it via nomotic inspect <agent-id>.

Accessing Verdict Details

verdict = runtime.evaluate(action, context)

print(verdict.verdict)            # The verdict enum
print(verdict.ucs)                # 0.0–1.0
print(verdict.tier)               # Which tier decided (1, 2, or 3)
print(verdict.reasoning)          # Human-readable explanation
print(verdict.vetoed_by)          # List of dimensions that vetoed
print(verdict.dimension_scores)   # All 14 dimension scores

Related

• UCS

• Three-Tier Evaluation

• Interrupt Authority