Agent Identity Security
Agent Identity Security is an add-on module for Prisma AIRS AI Gateway that gives
every enterprise AI agent a unique, verifiable identity with OAuth 2.1 client credentials,
fine-grained access policies over Model Context Protocol (MCP) server tools, and full
dual-attribution audit logging — integrated with your existing identity provider through
Cloud Identity Engine.
| Where Can I Use This? | What Do I Need? |
|---|
- Prisma AIRS — Agent Identity Security (US East region)
|
- Agent Identity Security module licensed and activated via Strata Cloud
Manager
- AI Gateway activated with at least one data plane instance running
- Enterprise IdP (Okta or Microsoft Entra ID) federated with CyberArk
Idira
- User directory synced to AI Gateway via Cloud Identity Engine (CIE)
|
AI agents reason, make decisions, and execute actions — both autonomously and on behalf
of human users. Today, most AI agents authenticate via inherited user credentials or
shared service accounts, leaving organizations with broad access risks and no clear trace
of which agent performed an action or whether it was authorized. Traditional Identity and
Access Management (IAM), Privileged Access Management (PAM), and Identity Governance and
Administration (IGA) frameworks depend on human interaction, while machine identity
solutions manage static workloads like service accounts. AI agents are fundamentally
different. They run at machine speed, operate non-deterministically, adapt to context,
and chain actions dynamically across systems. Agent Identity Security addresses these
challenges as an add-on module for Prisma AIRS, delivering unique, verifiable identities
and runtime controls to govern agent actions across all execution modes.
Agent Identity Security delivers unified identity, access control, and runtime governance
for enterprise AI agents. Agent Identity Security provides:
- A centralized catalog of every AI agent deployed in your enterprise. Each agent is
assigned a unique, system-generated Agent ID that persists across sessions and
systems.
- Owner assignment for every registered agent. Owners are mapped directly from your
connected identity provider (IdP via Idira/CIE) to ensure clear accountability for
lifecycle operations, policy modifications, and offboarding.
- Every registered agent is issued OAuth 2.1 client credentials to generate
short-lived, time-bound access tokens — eliminating the need for long-lived shared
secrets or human credentials.
- Fine-grained authorization policies that control which users, acting through which
agents, can call which Model Context Protocol (MCP) server tools. Policies are enforced
at runtime by the AI Gateway on every tool call.
- Every agent action is logged with both the agent identity and the user identity
(where applicable), giving you end-to-end traceability for investigation and
compliance.
System Architecture
Agent Identity Security integrates the Prisma AIRS AI Gateway with CyberArk's Secure
AI Identity Provider (Idira) and your existing federated human identity provider (IdP)
(such as Okta or Microsoft Entra ID).
Setup: AI Gateway needs to be connected to your enterprise Identity Provider
(IdP), such as Okta, using Cloud Identity Engine (CIE). Additionally, Idira also needs
to be federated with the same IdP to allow user login. When that is done, you must
register agents in Agent Identity Security, which results in the agent's OAuth 2.1
client credentials (client ID and secret) to be used later in the agent's OAuth 2.1
authorization code flow.
Agent and User Authentication: When an AI agent requests access to an MCP server
on behalf of a user, the AI Gateway authenticates both identities. The agent
authenticates using its registered OAuth 2.1 credentials, while the human user
authenticates through your connected enterprise identity provider to grant
authorization.
Outbound Authorization Handshake: Upon confirming that the requested access
complies with internal runtime policies, the AI Gateway requests user consent. Once
granted, it initiates the appropriate resource authentication flow, such as an additional
OAuth 2.1 exchange.
Prerequisites
Before configuring Agent Identity Security, ensure that you meet the following
prerequisites:
Your enterprise IDP (Okta or Microsoft Entra ID) must be connected to Idira (CyberArk
Identity Engine). This is a one-time configuration in the Idira tenant where you map the
user attribute (typically userPrincipalName) for identity federation. For more
information, see
Configure User Directory Synchronization in Idira.
Your user directory must be synced to the AI Gateway via Cloud Identity Engine (CIE).
This sync runs every 5 minutes, so newly created users may take up to 5 minutes to
become resolvable in policy enforcement. The AI Gateway must be activated and at least
one Gateway data plane instance registered and running. For more information, see
Synchronize User
Identity to AI Gateway via Cloud Identity Engine (CIE).
