Monitor governance activity
The Governance section is the dashboard your policies report into. It answers two operational questions at a glance: is anything waiting on a human, and is any policy acting where it should not.
Policies are authored in Policies, individual decisions are recorded in Decisions, and this dashboard is where both surface together. Learn more: Review policy decisions.
Read the headline metrics
-
Go to the Governance section under Trust & Governance in the navigation panel.
-
Select the time window. The metrics and panels below are scoped to it.
-
Review the three metrics.
Metric
What it counts
Pending approvals
Requests waiting for a reviewer decision. A number that keeps climbing means approvals are not being worked.
Policy approved
Human approvals completed in the window.
Total policy decisions
Every governance decision recorded in the window, whatever the result.
As a result, you can tell whether anything is stuck and whether your policies are recording decisions at all. A window with no decisions at all usually means nothing matched inside it rather than that a control has stopped working — widen the window before you conclude a policy is inactive.
Review pending approvals
The Approvals panel lists pending approvals from Human-in-the-Loop (HITL) policies and risky-tool confirmations. Open one to act on it, or open the paired decision to see the context that produced it.
Review recent decisions
The Recent policy decisions panel shows runtime and deployment-policy decisions across governed agents. Use it as the entry point into the full log when something looks wrong.
Find the agents that generate the most friction
The Decision hotspots by agent panel ranks agents by governance activity in the selected window.
Column | What it shows |
|---|---|
Agent runs | How much the agent ran, which puts the other columns in proportion. |
Policy triggers, Policy enforced | How often a policy matched, and how often it acted. |
Approval requested, Policy approved, Policy rejected | How much human review the agent generates, and how that review resolves. |
As a result, you can see which agents carry the most governance load. A high Policy enforced count against low Agent runs usually means the agent is mis-scoped rather than misbehaving. When the panel reports that no agents are standing out, no single agent is generating a disproportionate share of governance outcomes in the window.
The Top gated tools by approval volume panel answers the same question from the other side: which tools generate the approval traffic. Click a tool to open the decisions recorded for it. Policy hub at the top of the page opens the full list of policies.