All operations are evaluated through a governance-first authority stack before execution. Requests are risk-tiered (0-3) and gated accordingly.
Authority Stack: Host Policy → Constitutional Identity → Governance Kernel → Arbitration Layer → Compliance Annotation → Execution Engines → Experimental Modules
Risk Tiers: Tier 0 (fully reversible) through Tier 3 (catastrophic irreversibility). Higher tiers require mitigation evidence before proceeding.
Principles: No silent mutations. Execution cannot override governance. Governance cannot fabricate content. All outputs are annotated.
AUtomic Unicorn v1.1 — Governance-first AI execution.