Control Agent
Continuously evaluates every AI agent's identities, permissions, and access against your organization's security policies, preventing excessive permissions and policy violations.
What the Control Agent does inside Wing.
Continuously verify that AI agents comply with your organization's security policies: identity policies, access policies, permission assignments, and governance requirements.
Ensure every AI agent has the appropriate level of access. Monitor roles, permissions, identity assignments, privileged access, and authentication methods.
Identify excessive permissions, overprivileged agents, unused access rights, and opportunities to right-size AI agent access.
Detect when AI agents fall outside approved governance: configuration drift, unauthorized permission changes, and new policy violations.
Where the Control Agent reads and writes.
- Policy decisions are continuously validated, explainable, and auditable.
- Surfaces permission drift and excessive privileges before they create security incidents.
- Versioned policy snapshots with rollback support.
- Provides the Enforcement Agent context to act only on confirmed violations.
00:40:01 control.evaluate ok policy=allow action=read 00:40:14 control.evaluate HOLD reason=boundary.violation 00:40:29 control.drift WARN agent=copilot scope=repo+1 00:40:47 control.policy.publish v=2024.06.12 approved=sec-board