Live network-wide functionality

AI agent governance records

External client, policy profile, and execution-log records across the ecosystem.

empty

Visible records

Live Sanity records returned for this workstream.

Diagnostics

Superadmin data is network-wide and intentionally independent of the current domain shortcut.

Open canonical Studio ↗Manage network records in Sanity Studio.Configured Studio host ↗External Studio host. Check the functional action center status before relying on it for editor work.Open tenant registry →Create, search, and verify tenant records.Open operations →Review credential and deployment status.

Live records

0 returned · values are scoped and secret-safe

No live records returned for this workstream. This is a true empty state only when the status is empty; otherwise follow the diagnostics above.

Functional action center

External AI-agent governance

This module uses authenticated dashboard workflows and diagnostic envelopes instead of placeholder copy. It can read, create, moderate, or launch the correct provider surface depending on this workstream's guardrails.

empty

Deny-by-default access

Agent gateway status

Loading live provider state…

POST dispatch requires a bearer token. The dashboard intentionally does not collect or store that secret; operators should put it in provider secret storage and use scoped automation clients.

Audit scope

Visible agent records

No records match this filter. Empty means empty, not magic. Very enterprise.

← Back to superadmin dashboard

Network AI governance

External AI-agent governance

Register external agent clients, policy tiers, scope templates, MCP/API allowlists, denial rules, budgets, kill-switches, and logs.

policy gated

Purpose

This surface controls agent access across the entire platform and keeps automation least-privilege, draft-safe, rate-limited, and auditable.

Capabilities

Agent client registry
Policy tier templates L0-L3
Capability allow/deny rules
Rate/budget controls and kill-switch
Execution logs and denial review

Operational workflows

Register approved external clients.
Assign tenant/network scopes and tiers.
Review denials and suspicious activity.
Emergency-disable agent traffic if needed.

Guardrails

Default deny-by-default.
No cross-tenant access without explicit scope.
No provider secrets exposed through APIs/MCP.

API/schema contracts

These are technical contract references, not dashboard navigation. Protected APIs require authentication and may return JSON envelopes such as unauthenticated when opened directly.

/api/agent/gatewayintegrationClient schemaagentPolicyProfile schemaagentExecutionLog schema

Use the dashboard workstream pages for human workflows. Use these contracts only from authenticated clients, tests, cron jobs, or approved agent integrations.

Next implementation actions

Add client/profile management UI.
Add kill-switch control with confirmation/audit.
Add budget/rate-limit counters.