feature · r2-13-governance-policy
C
5 tags · 32 out · 32 in · 3 artifacts · 0 gaps · 1 corpus docs
Enterprise agent governance in 2026 converges on one architecture: probabilistic proposal → deterministic policy enforcement → append-only audit, with human oversight modeled as a finite, fatiguing resource — not an infinite safety net. Frameworks supply pause primitives (interrupt()); they do not supply approval-policy quality. That gap is now measurable (Headroom) and empirically visible (habituation study).
This topic was frozen in the corporate agentic OS research corpus (wave 10) to answer architecture and measurement questions that sit under the locked loop (observe→understand→predict→intervene→resolve→verify→learn). Stage-1 criticality: C. Research/Spec/Platform readiness: R5/S5/P3.
Graph edges for this feature — technologies, category lens, companies, DM.
features/r2-13-governance-policy/README.md
No open gaps for this entity.
T13 — Governance, Policy Gates, Trust · ok
features/r2-13-governance-policy/README.md
Research path: features/r2-13-governance-policy