If you’re building multi-agent AI systems, you need to prevent authorization scope from silently expanding as agents delegate tasks through multi-hop chains. Without proper controls, an agent can potentially act beyond what the originating user authorized, even when role-based access control (RBAC) policies are in place. The OWASP Top 10 for Agentic Applications classifies this risk as ASI03: Identity & Privilege Abuse.

This post shows you how to address the potential risk using a three-layer policy model built with Cedar, an open source authorization policy language, deployed on Amazon Web Services (AWS). The reference implementation uses OAuth 2.0 for authentication and Cedar for authorization. A trusted identity provider authenticates the originating user, then Cedar policies enforce authorization across three layers using verified token claims.

To enforce authorization at each hop in a multi-agent delegation chain, the reference implementation uses two AWS Lambda functions in sequence. A Model Context Protocol (MCP) adapter Lambda function normalizes inbound requests and cryptographically signs the originating user context. This prevents downstream tampering. A Cedar evaluator Lambda function evaluates three independent policy layers sequentially, halting on the first deny.

Table 1: Three-layer Cedar policy evaluation model

Cedar evaluates authorization but doesn’t establish identity. Before Cedar can evaluate context.originating_user.role or context.originating_user.mfa_verified, a trusted authentication layer must establish the user’s identity and produce verifiable claims. Steps 1–3 handle authentication; steps 4–10 handle authorization. The architecture shown in Figure 1 is described in the following lists:

1. The originating user authenticates with an OIDC-compliant identity provider (in this reference implementation, Amazon Cognito with TOTP multi-factor authentication (MFA)). The identity provider (IdP) issues a signed JSON Web Token (JWT) containing claims such as sub, role, amr (authentication methods), and session_id.
2. Amazon Cognito returns the signed JWT to the user.
3. The user passes the JWT and task request to the AI agent (MCP client). The agent carries the originating user context in the MCP _meta envelope.

4. The AI agent sends a Model Context Protocol (MCP) request to AWS WAF, which filters using CommonRuleSet, SQLiRuleSet, rate limiting, and body size constraints.
5. Amazon API Gateway (with Amazon Cognito authorizer) verifies the JWT signature against the user pool’s public keys and rejects invalid or expired tokens. Valid requests are forwarded to the MCP protocol adapter Lambda function, which applies Amazon Bedrock Guardrails content filtering.
6. The adapter extracts verified claims from the token and maps them to Cedar context attributes:
   1. JWT role claim : context.originating_user.role
   2. JWT amr includes MFA method: context.originating_user.mfa_verified = true
   3. JWT sub: context.originating_user.user_id
   4. JWT sid: context.originating_user.session_id
   5. JWT amr claim: context.originating_user.authentication_method

   The adapter then computes an HMAC-SHA256 signature over the user context (user_id, role, mfa_verified, authentication_method, and session_id in canonical order) using a key from AWS Secrets Manager.

7. The adapter constructs a signed request envelope and invokes the Cedar evaluator Lambda function.
8. The evaluator verifies the HMAC-SHA256 signature, retrieves L2 and L3 Cedar policies from Amazon Verified Permissions, and evaluates all three layers (L1, L2, and L3), halting on the first deny.
9. The evaluator emits an Open Cybersecurity Schema Framework (OCSF) 99001 audit event to Amazon CloudWatch Logs. Failed emissions fall back to an Amazon Simple Queue Service (Amazon SQS) dead-letter queue (DLQ).
10. Amazon CloudWatch dashboards and alarms monitor evaluation latency, deny rates, and DLQ depth. Alarm notifications route through Amazon Simple Notification Service (Amazon SNS).

Two mechanisms work together to protect identity across hops:

• Hash-based Message Authentication Code (HMAC-SHA256) ensures integrity and authenticity. Every downstream evaluator verifies this signature before trusting the context.
• OAuth 2.0 Token Exchange (RFC 8693) sets delegation scope using the on-behalf-of (OBO) pattern. When the orchestrator delegates to a downstream agent (data-bot), it exchanges the original token for a scoped OBO token that records who’s acting on behalf of whom and with what authority. The Cedar policies (detailed in Step 2: Three-layer policies) then check whether that scoped delegation is permitted and verify the originating user claims carried in the OBO token. Token exchange limits each downstream agent to only the delegated task’s scope instead of passing through the full original token. For enterprise deployments, use token exchange alongside HMAC. OAuth tracks who is acting on behalf of whom and with what scope. HMAC verifies that the context hasn’t been tampered with and came from a trusted source.

The following prerequisites are needed to deploy the reference implementation. Before you begin, clone the repository:

git clone https://github.com/aws-samples/sample-cedar-agentic-ai-authorization.git
cd sample-cedar-agentic-ai-authorization

Verify that you have the following:

• An AWS account with AWS Cloud Development Kit (AWS CDK) bootstrapped in the target AWS Region
• Python 3.12 or later, and Node.js 20 or later (for CDK command line interface (CLI))
• AWS CDK CLI (npm install -g aws-cdk)
• AWS credentials with scoped AWS Identity and Access Management (IAM) policy permissions for Lambda, API Gateway, Verified Permissions, Secrets Manager, Amazon Virtual Private Cloud (Amazon VPC), and CloudWatch

In this walkthrough, you define the Cedar entity schema and policies, deploy the infrastructure with AWS CDK, and integrate your identity provider.

In this step, you define a schema with two entity types (Agent and Tool) and two actions (invoke_tool and delegate_task) in the AgentAuthz namespace. Notice that there is no User entity. Instead, you carry the originating user’s identity in the evaluation context record, which is a structured data object passed alongside each authorization request.

{
  "AgentAuthz": {
    "entityTypes": {
      "Agent": {
        "shape": {
          "type": "Record",
          "attributes": {
            "trust_level": { "type": "Long", "required": true },
            "namespace": { "type": "String", "required": true },
            "registered_capabilities": {
              "type": "Set", "element": { "type": "String" }, "required": true
            },
            "lifecycle_stage": { "type": "String", "required": true }
          }
        }
      },
      "Tool": {
        "shape": {
          "type": "Record",
          "attributes": {
            "namespace": { "type": "String", "required": true },
            "risk_level": { "type": "String", "required": true }
          }
        }
      }
    },
    "actions": {
      "invoke_tool": {
        "appliesTo": { "principalTypes": ["Agent"], "resourceTypes": ["Tool"] }
      },
      "delegate_task": {
        "appliesTo": { "principalTypes": ["Agent"], "resourceTypes": ["Agent"] }
      }
    }
  }
}

This schema is deployed to an Amazon Verified Permissions policy store by the VerifiedPermissionsStack CDK stack. In the reference implementation, the schema file is located at cedar-entity-schema.json.

Agent topology and attributes

The following tables show the agents and tools registered in this reference implementation, along with the attributes the Cedar evaluator function retrieves from the entity store. The test scenarios that follow trace requests through this topology.

Table 2: Agent attributes

Table 3: Tool attributes

The orchestrator can delegate to both data-bot and finance-agent. Each agent can only invoke tools within its registered capabilities. The test scenarios below trace requests through these delegation paths.

The following policies are deployed to the same Verified Permissions policy store. In the reference implementation, policy files are located under cedar/policies/ organized by layer: layer1-agent-to-tool/, layer2-agent-to-agent/, and layer3-originating-user-auth/.

Layer 1 (agent-to-tool): This policy permits the finance-agent to invoke the process_payment tool only when three conditions are met: the agent’s trust score is at least 3, it belongs to the payments namespace, and it’s deployed in the production lifecycle stage. If any condition fails, the request is denied. The agent’s trust_level, namespace, and lifecycle_stage aren’t self-reported in a production deployment. Instead, the evaluator retrieves these attributes from the Verified Permissions entity store using the agent_id as a lookup key.

Important: The reference implementation accepts these values from the request payload for simplicity. Production deployments must validate agent attributes against an authoritative source to prevent a compromised agent from escalating its own trust.

The trust_level attribute uses a 1–5 integer scale that represents an agent’s verified maturity: 1 for newly registered and untested agents, 3 for agents that have passed integration testing and security review, and 5 for agents with a proven production track record. Organizations assign trust levels through their agent promotion pipeline, not through self-declaration. The lifecycle_stage attribute (development, staging, production) prevents pre-production agents from invoking production tools, even if they have the correct namespace and trust score.

// L1-001: Finance agent can invoke payment tools
permit(
  principal == AgentAuthz::Agent::"finance-agent",
  action == AgentAuthz::Action::"invoke_tool",
  resource == AgentAuthz::Tool::"process_payment"
) when {
  principal.trust_level >= 3 &&
  principal.namespace == "payments" &&
  principal.lifecycle_stage == "production"
};

Layer 2 (agent-to-agent delegation) enforces depth limits and capability constraints. The orchestrator agent delegates tasks to data-bot only when the delegation chain is three hops or fewer and the requested capabilities are a subset of data-bot’s registered capabilities. A separate forbid policy (L2-004) enforces a hard system-wide limit of five hops regardless of which agents are involved.

// L2-002: Orchestrator can delegate to data agent
permit(
  principal == AgentAuthz::Agent::"orchestrator",
  action == AgentAuthz::Action::"delegate_task",
  resource == AgentAuthz::Agent::"data-bot"
) when {
  context.delegation_depth <= 3 &&
  context.target_capabilities.containsAll(context.requested_capabilities)
};

Layer 3 (originating user authorization) keeps the agent as the principal, but the policy evaluates context.originating_user to validate the human who initiated the request. data-bot invokes the delete_records tool only when the originating user has the admin role, has verified MFA, and the delegation chain is at most two hops deep. Without this layer, an agent with the right capabilities could invoke destructive tools regardless of who initiated the request.

// L3-001: High-risk tool (delete_records) requires admin + MFA
permit(
  principal == AgentAuthz::Agent::"data-bot",
  action == AgentAuthz::Action::"invoke_tool",
  resource == AgentAuthz::Tool::"delete_records"
) when {
  context.originating_user.role == "admin" &&
  context.originating_user.mfa_verified == true &&
  context.delegation_depth <= 2
};

Key design point: The principal remains the agent, not a user entity. The user’s role and MFA status are checked through context attributes, keeping the schema to two entity types and two actions.

The reference implementation uses Amazon Cognito with TOTP MFA, but most OIDC-compliant providers (Okta, Microsoft Entra ID, Auth0, or AWS IAM Identity Center) work with this pattern. The authentication-to-signing flow is described in the preceding Authentication before authorization section. To use a different IdP, replace the Cognito authorizer on API Gateway with a Lambda or JWT authorizer for your IdP’s issuer URL. Cedar policies remain unchanged.

The reference implementation deploys five CloudFormation stacks: KmsStack, VerifiedPermissionsStack, LambdaStack, SecurityLakeStack, and MonitoringStack. The following commands deploy the stacks in dependency order:

cdk deploy KmsStack -c account_id=YOUR_ACCOUNT_ID -c guardrail_id=YOUR_GUARDRAIL_ID
cdk deploy VerifiedPermissionsStack -c account_id=YOUR_ACCOUNT_ID -c guardrail_id=YOUR_GUARDRAIL_ID
cdk deploy LambdaStack -c account_id=YOUR_ACCOUNT_ID -c guardrail_id=YOUR_GUARDRAIL_ID
cdk deploy SecurityLakeStack -c account_id=YOUR_ACCOUNT_ID -c guardrail_id=YOUR_GUARDRAIL_ID
cdk deploy MonitoringStack -c account_id=YOUR_ACCOUNT_ID -c guardrail_id=YOUR_GUARDRAIL_ID

Three end-to-end scenarios validate the evaluation model across different user roles, MFA states, and delegation depths. To run the tests:

1. Set the API endpoint from the deployment output:

export API_ENDPOINT=$(aws cloudformation describe-stacks --stack-name LambdaStack \
  --query "Stacks[0].Outputs[?OutputKey=='ApiEndpoint'].OutputValue" --output text)

2. Run the end-to-end tests:

.venv/bin/python -m pytest tests/e2e/ -v -s

The end-to-end tests cover the three scenarios described in the following sections. Each test sends a request through the deployed API and validates the per-layer authorization decisions.

A support-role user (no MFA) requests record deletion through orchestrator and data-bot.

Without Layer 3, this request would have been permitted based on agent capabilities alone, demonstrating why originating user authorization is essential.

An admin user with MFA requests the same operation through the same chain.

An admin with MFA requests the same operation, but the delegation chain has six hops. This scenario tests the Layer 2 depth constraint independently of user authorization.

Even an authorized admin can’t bypass the delegation depth constraint.

The AWS Office of the CISO published Four security principles for agentic AI systems. The following table shows how this solution maps to each principle.

Each evaluation produces an OCSF 99001 audit event with request ID, user identity, delegation chain, per-layer decisions, and latency.

The following table maps this implementation to NIST SP 800-53 Rev. 5 controls. Customers are responsible for evaluating whether it meets their compliance requirements.

Deploy the Cedar policy store in a central security account and use cross-account IAM roles for workload accounts to call verifiedpermissions:IsAuthorized. Use AWS Organizations service control policies (SCPs) to prevent workload accounts from creating their own policy stores. For standardizing user identity attributes across the organization, consider IAM Identity Center or a centralized OIDC provider that issues consistent claims to your workload accounts. This helps ensure that the context.originating_user attributes are uniform across accounts and agents.

For production deployments, consider extending this pattern with human-in-the-loop escalation for borderline denials, multi-tenant Cedar policy isolation, and Amazon Simple Storage Service (Amazon S3)-backed dynamic policy hot-reload for emergency tool shutdowns.

To avoid ongoing charges, delete the deployed resources:

cdk destroy MonitoringStack SecurityLakeStack
cdk destroy LambdaStack
cdk destroy VerifiedPermissionsStack
cdk destroy KmsStack
aws logs delete-log-group --log-group-name /cedar-evaluator/audit  # if RETAIN policy

Multi-agent AI systems need authorization boundaries at every delegation hop. The three-layer Cedar policy model with OAuth 2.0 authentication provides that protection while maintaining least-privilege access. Combining a trusted IdP (AuthN) with Cedar policy evaluation (AuthZ) creates an authorization boundary around each tool invocation, verifying agent capability (L1), delegation path (L2), and originating user authority (L3). The pattern works with an OIDC-compliant IdP and a compute platform that can call Amazon Verified Permissions. Clone the reference implementation and adapt the Cedar policies to your organization’s requirements. For more information, see the Cedar policy language documentation and the Amazon Verified Permissions User Guide.

• Cedar policy language documentation
• Amazon Verified Permissions User Guide
• OWASP Top 10 for Agentic Applications
• Four security principles for agentic AI systems
• OAuth 2.0 Token Exchange (RFC 8693)
• NIST AI Agent Standards Initiative
• NIST SP 800-53 Rev. 5
• OCSF schema documentation

If you have feedback about this post, submit comments in the Comments section below.

Modern web applications require robust security controls to protect user data and application resources. Authentication and authorization are two fundamental pillars of application security that answer critical questions: Who are you? and What are you allowed to do? Implementing these controls correctly can be challenging for developers, especially when building data-intensive applications with frameworks like Streamlit (an open-source Python framework for building interactive web applications) or when requiring fine-grained access control. Key challenges include protecting access to application resources, implementing application identity with multi-factor authentication (MFA), and implementing usage-based controls.

In this post, you will learn how to build fine-grained access controls for a sample Streamlit application using Amazon Cognito for authentication and Amazon Verified Permissions with Cedar policies for authorization. This architecture provides enterprise-grade security with minimal development effort, so you can focus on your application’s core functionality. You will learn how to reduce development time for secure applications, implement enterprise-grade authentication, through proper access management, and scale security with growing user bases.

Security architecture overview

The reference architecture follows a layered security design with four key components; separating identity verification, authorization evaluation, application logic, and enforcement boundaries. By assigning clear responsibilities to each layer, the architecture limits blast radius and ensures that a failure in any single control does not compromise the overall system.

• Authentication layer: Amazon Cognito handles user authentication with secure credential validation and JSON web tokens (JWTs). It provides built-in password policies, account lockout protection, and session management.
• Authorization layer: Verified Permissions uses the Cedar policy engine to evaluate fine-grained access requests based on centrally stored policies.
• Application layer: The Streamlit frontend integrates with both services, managing user sessions and enforcing access controls in the user interface.
• Security boundaries: Multiple layers of security controls protect against unauthorized access, privilege escalation, authentication verification, authorization checks, and input validation.

This separation of concerns enables authentication and authorization to function as complementary security controls, following defense-in-depth principles. Figure 1 illustrates the end-to-end authentication and authorization workflow, showing how a user’s sign-in request flows through Amazon Cognito for identity verification, then through Verified Permissions for Cedar policy-based access decisions, before the application enforces the result.

Figure 1: Solution architecture and workflow

The following workflow demonstrates how the three architecture layers work together: the authentication layer (steps 1–3) handles identity verification using Amazon Cognito, the authorization layer (steps 4–6) evaluates Cedar policies using Verified Permissions, and the application layer (steps 7–8) enforces the decision in Streamlit.

1. The user sends a sign-in request, which is submitted through Streamlit
2. The request is authenticated by Amazon Cognito
3. An access token is sent back to Streamlit
4. An authorization request is sent to Verified Permissions
5. The Cedar policy engine evaluates the request
6. A decision is sent back by the policy engine
7. The instruction to allow or deny is sent back to Streamlit
8. If the instruction is to allow, access is provided

Understanding authorization with Cedar

While authentication establishes user identity, authorization determines what actions users can perform. Verified Permissions provides a scalable authorization service based on Cedar, a policy language specifically designed for fine-grained access control.

Cedar policies follow a structured format that defines who can perform which actions on what resources. Let’s examine the anatomy of a Cedar policy:

permit(
    principal == ?principal,
    action == application::Action::"ViewGrade",
    resource == ?resource
) when {
    principal has role == "Student" &&
    resource.student == principal.entityId
};

Policy components

• Effect: permitor forbid determines whether the policy allows or denies access
• Principal: The entity (user) making the request, represented by ?principal as a variable
• Action: The operation being performed, scoped to your application namespace
• Resource: The target of the action, also represented as a variable
• Conditions: The when clause contains logical expressions that must evaluate to true

Advanced Cedar policy patterns

This section describes commonly used Cedar policy patterns for implementing fine-grained authorization with Amazon Verified Permissions. The examples illustrate how to model ownership, role-based access, hierarchical permissions, and administrative controls in real-world applications

Resource ownership control

This pattern helps ensure that users can only access resources they own:

permit(
    principal == ?principal,
    action == application::Action::"ViewGrade",
    resource == ?resource
) when {
    principal has role == "Student" &&
    resource.student == principal.entityId
};

What it does – This policy allows students to view only their own grades by:

• Checking that the user has the Student role
• Verifying that the grade resource’s student attribute matches the student’s entityId
• Preventing students from accessing other students’ grades while allowing access to their own academic performance

Role-based access with resource type

This pattern grants access based on role and resource type:

permit(
    principal == ?principal,
    action == application::Action::"EditCourse",
    resource == ?resource
) when {
    principal has role == "Faculty" &&
    resource has resourceType == "Course" &&
    resource.instructor == principal.entityId
};

What it does – This policy allows faculty members to edit courses they teach by:

• Verifying the user has the Faculty role
• Confirming the resource is of type Course
• Verifying that the course’s instructor attribute matches the faculty member’s entityId
• Restricting faculty to modify only their own courses, not courses taught by other instructors

Hierarchical authorization

This pattern allows department heads to manage faculty in their department:

permit(
    principal == ?principal,
    action == application::Action::"ManageFaculty",
    resource == ?resource
) when {
    principal has role == "DepartmentHead" &&
    resource has role == "Faculty" &&
    resource.department == principal.department
};

What it does – This policy implements departmental hierarchy controls by:

• Requiring the user to be a DepartmentHead
• Verifying the resource is a faculty member
• Matching the faculty member’s department with the department head’s department
• Preventing department heads from managing faculty in other departments

Administrative override

This pattern provides emergency access with proper justification:

permit(
    principal == ?principal,
    action == ?action,
    resource == ?resource
) when {
    principal has role == "Administrator" &&
    context has emergencyAccess == true &&
    context has justification
};

What it does – This policy provides emergency access capabilities by:

• Allowing administrators to perform any action on any resource
• Requiring an emergency access flag to be set to true
• Requiring a justification for emergency access
• Supporting accountability through required documentation while enabling emergency operations

Cedar policy evaluation flow

Understanding how policies are evaluated helps design effective authorization systems. Figure 2 shows a common evaluation pattern for an academic scenario

Note: A policy match evaluates to the policy’s effect (permit or forbid). Forbid policies take precedence: if any forbid policy matches, access is denied regardless of permit policies.

Figure 2: Policy evaluation process

The policy evaluation process follows these steps:

1. User attempts to access a protected resource
2. Application sends an authorization request to Verified Permissions
3. Verified Permissions retrieves applicable Cedar policies from the policy store
4. The Cedar policy engine evaluates each policy against the request
5. If any forbid policy matches, access is denied immediately
6. If any permit policy matches and no forbid policies match, access is allowed
7. If no policies match, access is denied by default
8. The evaluation result (ALLOW or DENY) is returned to the application
9. Application enforces the authorization decision

Cedar policy language

Cedar is an Amazon open source policy language designed for fine-grained authorization. Every policy defines who (principal) can perform what action on which resource under what conditions, as shown in Figure 3.

Figure 3: Cedar policy definitions

Policy interaction

The following table shows how different policies interact in complex scenarios where multiple policies could apply:

Legend:

• Permit – Policy allows access
• N/A – Policy doesn’t apply
• Override – Emergency admin access

The preceding table shows how each role’s policy applies to different scenarios, with admin access having override capabilities across most situations except for emergency admin access where it’s the primary permit authority. The Override column specifically indicates that the administrator’s emergency access policy can supersede other role-specific policies, but only when the emergencyAccess context flag is explicitly set and a justification is provided. This is not an automatic override.

Policy optimization tips:

• Order conditions by likelihood of success – Place the most frequently true conditions first in your when clause to enable short-circuit evaluation. For example, check role before resource ownership, because role mismatches are caught earlier. See Cedar best practices.
• Use indexed attributes for faster lookups – Use entity attributes that Verified Permissions indexes natively (entityId, role, resource type) as primary conditions. Best practices for designing an authorization model
• Cache policy evaluations when appropriate
• Monitor evaluation metrics and performance

Real-world application: Academic system

Consider an academic system with different user roles and their corresponding permissions:

Student: View own grades

• Policy helps ensure students can only access grade resources where they are listed as the student
• The policy verifies the student’s role and matches the resource owner to the principal’s entity ID

Faculty: Edit course content, manage grades

• Policy allows faculty to edit courses they teach
• Faculty can view and modify grades for students in their courses

Teaching assistant (TA): Grade management and course support

• Policy permits TAs to manage grades for courses they assist with
• Access is limited to specific courses assigned to the TA

Department head: Manage faculty assignments

• Policy allows department heads to manage faculty in their department
• Access is scoped to the department hierarchy

Administrator: System-wide access

• Policy provides emergency access with proper justification
• Administrative actions are logged and audited

Prerequisites

To implement the preceding Academic system application, you need an active AWS account, Python 3.8 or later, basic Streamlit knowledge, and AWS Identity and Access Management (IAM) permissions for Amazon Cognito and Verified Permissions.

Run the sample and extend the solution

1. Download the code base: Start by downloading the code base from the avp streamlit samples repository
2. Set up your development environment: Install the AWS SDK for Python (boto3) and configure your AWS credentials.
   • Install the AWS SDK for Python:

     pip install boto3
     

   • Log in to your AWS account:

     aws login --region $REGION

   • Verify that your AWS Command Line Interface (AWS CLI), Python, and dependencies are correctly configured.

     ./verify-setup.sh

3. Create your AWS resources: Use the AWS Management Console or infrastructure as code (IaC) tools to provision your Amazon Cognito user pool and Verified Permissions policy store.

   ./deploy-demo-environment.sh
   Do you want to start the demo now? (Y/N): Y

   This provisions an Amazon Cognito user pool, a Verified Permissions policy store, and any sample resources needed for the demo.

4. Verify the login screen:
   
   

   Figure 4: Verify login credentials

5. Demo walkthrough and shut down: Interact with the demo and test the policies and features. When you’re ready to exit, press Ctrl+C to shut down and stop.
6. Define your Cedar policies: Start with basic policies and gradually add complexity as you understand the evaluation model.
7. Implement authentication: Integrate Amazon Cognito authentication into your application with proper error handling.
8. Add authorization checks: Implement authorization checks at critical access points in your application. For authentication, implement proper error handling for expired tokens, failed MFA challenges, and account lockouts. Use the Amazon Cognito built-in token refresh flow. For authorization, place Verified Permissions checks at every API endpoint and UI component that accesses protected resources.
9. Test thoroughly: Create test scenarios for each user role and permission combination.
10. Monitor and iterate: Set up AWS CloudTrail logging and Amazon CloudWatch alarms to monitor your security controls and refine them based on real-world usage.

Security best practices

When implementing this architecture, follow these best practices to support security:

• Layer your security controls: Use both authentication and authorization as complementary controls rather than relying on a single mechanism.
• Follow least privilege principles: Grant only the permissions needed for specific user roles. Start with minimal permissions and add more as needed.
• Implement proper session management: Set appropriate token expiration and refresh policies. Amazon Cognito handles much of this automatically, but you should configure timeouts based on your security requirements.
• Validate all inputs: Sanitize user inputs to prevent injection attacks. Don’t rely on client-side validation alone.
• Monitor authentication events: Set up logging and alerts for suspicious activities such as repeated failed login attempts or unusual access patterns.
• Conduct regular security reviews: Periodically audit your policies and security configurations to verify they still meet your requirements and follow current best practices.
• Implement secure error handling: Avoid information disclosure through error messages. Provide helpful feedback to users without revealing system details that could aid attackers.

Conclusion

Implementing proper authentication and authorization is critical for application security. By using Amazon Cognito and Amazon Verified Permissions, you can build robust security controls without complex custom code. Through this approach, you can implement enterprise-grade authentication with minimal effort, define and enforce fine-grained authorization policies, scale your security controls as your application grows, and centrally manage and audit security policies.

To get started with your implementation, create your AWS resources including an Amazon Cognito user pool and Verified Permissions policy store. Define your Cedar policies based on your application’s access requirements. Integrate authentication and authorization checks into your application flow. Test thoroughly with different user roles and access scenarios. Finally, monitor and refine your security controls based on usage patterns.

For additional resources, check out the Amazon Cognito documentation and Amazon Verified Permissions documentation.

If you have feedback about this post, submit comments in the Comments section below.

Agents have agency: they adapt and find multiple ways to solve problems. This autonomy creates a fundamental security challenge: the large language model (LLM) at the heart of the agent is non-deterministic, and its decisions can’t be predicted or guaranteed in advance. It can hallucinate harmful actions with complete confidence. It’s vulnerable to prompt injection attacks, where adversaries inject malicious commands through tool responses or user inputs. LLMs don’t robustly differentiate between commands and data, everything is only tokens. For these reasons, if you want defense in depth, you must treat the LLM as an untrusted actor from a security point of view.

The insight is that the LLM can’t affect the external world directly: it has to go through an orchestrator that invokes tools based on the LLM’s output. This is precisely where the controls must be applied. What you need at this boundary is authorization: a decision about whether each tool invocation should be allowed and under what conditions. Consider a customer service agent for an online retailer. Without proper controls, it could process refunds that exceed authorized limits, apply discounts to product categories that should be excluded, or look up one customer’s data while handling another customer’s session.

If you control agents’ access to tools, you can establish a safety envelope within which the agent can operate freely. This differs from two common but unsatisfactory approaches:

• Creating hard-coded workflows eliminates uncertainty, but by itself defeats the purpose of using an LLM as the brain of the agent, because you’ve built a traditional application with an LLM interface. And even with this restriction, using LLM outputs at any step can open up the same risks. While it’s a useful technique for well-understood workflows, it’s not sufficient for agents that need to adapt.
• Human-in-the-loop provides a safety net for critical operations, and it will always have a role. But relying on it as the main control mechanism sacrifices autonomy and can lead to approval fatigue.

You need agents that are safe and autonomous. This requires an auditable, deterministic enforcement layer that sits outside the agent and tools. Why outside? Because the LLM’s plan is the thing you can’t trust—it can’t be responsible for enforcing its own constraints. Controls at the LLM layer—such as system prompts and training-time alignment—can be bypassed by prompt injection or hallucination. Hard-coded checks in agent or tool code are more robust, but become difficult to audit and manage at scale, especially when security logic is scattered across many tools and services. Centralizing authorization outside both gives you a single checkpoint the LLM can’t circumvent; one that’s auditable and can be verified independently of the application code.

This is where AgentCore Policies come in. Amazon Bedrock AgentCore Gateway sits between the agent and the remote tools it calls. When you associate a Policy with a Gateway, it blocks everything by default. Policies selectively open this boundary by specifying which tool invocations are allowed and under what conditions. This enforcement applies to all tool traffic routed through the Gateway. For this approach to scale, it must be more straightforward to reason about the policies than about the agent’s behavior.

AgentCore policies are expressed in Cedar. Cedar is an open source authorization policy language developed by AWS that has recently joined the Cloud Native Computing Foundation (CNCF). Cedar was designed with exactly these properties: it’s purpose-built for authorization, readable by humans, and analyzable by machines using automated reasoning. This gives enterprises the ability to scale policy definition and enforcement to their AI agents.

Amazon Bedrock AgentCore provides the infrastructure to deploy and manage agents at scale. It includes AgentCore Runtime for hosting agents, AgentCore Gateway for managing how agents connect to tools using Model Context Protocol (MCP), and Policy in AgentCore. Policy intercepts all agent traffic through AgentCore gateways and evaluates each request against defined policies in the policy engine before allowing tool access. Cedar powers the policy layer.

AgentCore Policy uses Cedar and its mathematical analysis capabilities at several points in the AgentCore Gateway workflow: the Cedar authorization engine is used at policy evaluation and Cedar Analysis is used during policy authoring, and in the control plane.

Policy authoring: Developers can write Cedar policies directly or use natural language that gets translated to Cedar through a neuro-symbolic AI feedback loop. Neuro-symbolic AI combines machine learning’s flexibility with automated reasoning’s provable correctness. An LLM generates policies from natural language, while Cedar Analysis validates them using symbolic, mathematical reasoning. The following diagram illustrates this workflow:

Figure 1: Cedar policy generation workflow

An administrator specifies—in natural language—which MCP tools the agent can call and under what conditions. The neuro-symbolic feedback loop then formalizes this description into Cedar policies. Here’s how it works: first, the LLM translates the natural language into Cedar policies. These policies are then run through two stages of verification. In the first stage, AgentCore Policy uses a Cedar schema generator that takes the MCP tool descriptions and produces a Cedar schema. Cedar validates the policies against this schema, helping to ensure that they reference valid tools and parameters and ruling out whole classes of runtime errors. If validation passes, the second stage runs Cedar Analysis, which encodes each policy as a mathematical formula and detects issues like policies that grant or deny everything, or that contain impossible conditions. These mathematical proofs identify errors in the process of translating from the natural language description to Cedar policies, and guide corrections.

The neuro-symbolic feedback loop significantly improves the accuracy of the generated policies. This demonstrates the power of combining neural and symbolic approaches—the LLM provides creative translation from natural language, while automated reasoning provides rigorous validation.

Control plane: When attaching policies to an AgentCore Gateway, Cedar Analysis performs holistic analysis of the entire policy set. Instead of analyzing policies in isolation, it examines how they interact and their combined effect. This analysis identifies potential logical errors—such as conflicting or redundant policies—and detects whether the policy set produces unintended authorization outcomes. When Cedar Analysis detects these errors, the operation fails and returns a description of the issue, so the policy author can fix and retry. See the Formal analysis for policy verification section for examples of the checks.

MCP tool invocation enforcement: Each agent tool request made to the AgentCore gateway is evaluated against Cedar policies which determine whether the MCP tool invocation with the given arguments should be allowed. This creates the safety envelope while allowing the necessary bridges to enable the agent to perform its job.

MCP tool filtering: Cedar enables an additional layer of protection that operates before any tool invocation occurs. When an agent issues a list tools command, AgentCore Gateway uses Cedar’s partial evaluation capability to determine which actions would always be denied under the current policy set. Those actions are omitted from the list tool response. The agent and the underlying LLM never see those tool actions, eliminating an entire class of risk: the agent and LLM can’t attempt to invoke a tool it doesn’t know exists. This is a direct benefit of Cedar’s partial evaluation: the system can determine that certain tool actions are unreachable without needing to wait for an actual tool invocation attempt.

Natural language is too ambiguous for security-critical infrastructure, and general-purpose programming languages, like Python, are very expressive but too difficult to analyze. They can have unintended side effects, termination issues, and can be difficult to understand.

Cedar avoids these issues by excluding loops and stateful operations, so policy evaluation terminates in O(n) time in common cases. This bounded execution time means agents can make authorization decisions without disrupting user experience or workflow efficiency.

Cedar is straightforward to read. Regulatory compliance and security audits require policies that humans can understand and verify. Cedar policies read like structured natural language, making them accessible to security teams, compliance officers, and business stakeholders:

// Only allow bulk discounts for premium customers with sufficient quantity
permit (
  principal is AgentCore::OAuthUser,
  action == AgentCore::Action::"ApplyBulkDiscount",
  resource
)
when
{
  principal.hasTag("customer_tier") &&
  principal.getTag("customer_tier") == "Platinum" &&
  context.input.orderQuantity >= 50
}
unless
{
  context.input
    .productTypes
    .containsAny
    (
      ["limited_edition", "seasonal_specials"]
    )
};

Auditors without a technical background can understand this policy: “Allow bulk discounts for platinum customers who order at least 50 items, except for limited edition or seasonal special products.” The unless clause makes the exception clear, which is how business rules are typically expressed in natural language. Notice that this single policy constrains two different sources of data. The customer tier comes from a JSON Web Token (JWT) claim—it can’t be hallucinated or manipulated by the LLM. The tool inputs like order quantity and product types, however, originate from the LLM’s tool call. Cedar policies constrain these inputs to only allowed values, ensuring that even if the LLM produces unexpected arguments, the policy enforcement layer rejects them deterministically.

Cedar is the right choice because it’s fast, straightforward to read, and analyzable through automated reasoning. This analyzability is why you can reason about the safety envelope around agents that’s expressed as Cedar policies. As agentic systems grow the number of tools grows. Without proper tooling, policy management becomes intractable; policies can conflict, create security gaps, or produce unintended authorization outcomes.

In the rest of this section, we examine how Cedar’s analyzability directly addresses this challenge through its deterministic, mathematically sound analysis. Because Cedar analysis can reliably detect conflicts and logical errors across large policy sets it enables scalable policy management through neuro-symbolic AI.

Cedar policies can be encoded as mathematical formulas and analyzed using automated reasoning techniques through a symbolic encoder. This enables AgentCore Policy to provide sophisticated policy verification capabilities during policy authoring and beyond. AgentCore Policy uses this analysis when authoring or attaching policies to detect possible logical errors, such as conflicting or redundant policies. Policy analysis, including policy comparison is available as an open source CLI tool. Next, we will take a look at some concrete examples of these checks.

Detecting logical errors in policies: Cedar Analysis can detect when policies contain logical errors. For example, the following policy has contradictory constraints that mean it can’t allow any request: the customer tier can’t be both gold and platinum at the same time. The intention was to use an || instead of &&, a mistake that can be made by both humans and AI systems that author policies.

// This policy cannot allow any requests due to logical errors
permit (
  principal is AgentCore::OAuthUser,
  action == AgentCore::Action::"ProcessRefund",
  resource
)
when
{
  principal.hasTag("customer_tier") &&
  principal.getTag("customer_tier") == "Gold" &&
  principal.getTag("customer_tier") == "Platinum"
}
unless { context.input.refundAmount > 1000 };

Similarly, Cedar Analysis can detect policies that always allow a given action, usually an indication of an overly permissive policy. For example, the following policy will allow all ApplyBulkDiscount requests because any order quantity will either be greater than or equal to 100 or less than 100.

// This policy allows all ApplyBulkDiscount requests
permit (
  principal is AgentCore::OAuthUser,
  action == AgentCore::Action::"ApplyBulkDiscount",
  resource
)
when
{
  context.input.orderQuantity >= 100 ||
  context.input.orderQuantity < 100 ||
  (principal.hasTag("customer_tier") &&
   principal.getTag("customer_tier") == "Platinum")
};

Detecting such logical errors isn’t easy for humans, and can’t be done by pattern matching: you need the formal rigor of mathematical analysis, which is exactly what Cedar Analysis does.

Detecting policy conflicts: Cedar Analysis can also analyze the entire policy set to detect inconsistencies between different individual policies:

// These policies conflict - Analysis will detect the subtle issue
permit (
  principal is AgentCore::OAuthUser,
  action == AgentCore::Action::"ProcessRefund",
  resource
)
when
{
  principal.hasTag("customer_tier") &&
  principal.getTag("customer_tier") == "Gold" &&
  context.input.refundAmount < 100
};

forbid (
  principal is AgentCore::OAuthUser,
  action == AgentCore::Action::"ProcessRefund",
  resource
)
when
{
  principal.hasTag("customer_tier") &&
  ["Gold", "Platinum"].contains(principal.getTag("customer_tier")) &&
  context.input.refundAmount < 500
};

The permit policy allows gold customers to process refunds less than $100, while the forbid policy blocks gold customers (and platinum customers) from processing refunds less than $500. Because forbid overrides permit in Cedar, the forbid policy would block all gold customer refunds despite the permit policy.

Comparing policy changes: When updating policies, Cedar Analysis can also determine the exact impact of a change. Consider the following update to the unless clause (the policy lines with + have been added and those with - have been removed): we now block ApplyBulkDiscount only when the product type is limited_edition and the quantity exceeds 200.

 permit (
   principal is AgentCore::OAuthUser,
   action == AgentCore::Action::"ProcessRefund",
   resource
 )
 when
 {
   context.input.refundAmount < 500
 };
 
 permit (
   principal is AgentCore::OAuthUser,
   action == AgentCore::Action::"ApplyBulkDiscount",
   resource
 )
 when
 {
   context.input.orderQuantity >= 50
 }
 unless
 {
-  context.input.productTypes.containsAny(["limited_edition"])
+  context.input.productTypes.containsAny(["limited_edition"]) &&
+  context.input.orderQuantity > 200
 };

At first glance, adding a condition to the unless clause might seem more restrictive. In fact, it’s the opposite: narrowing when the unless applies means the permit now covers more requests. For example, an order of 73 units of a limited_edition product would have been blocked before but is now allowed. Cedar Analysis can automatically detect this and generates the following table showing the difference in permissiveness between the original policy set and the updated one:

In the preceding example, the analysis tells us that the updated policy allows allows exactly the same ProcessRefund requests, but allows more ApplyBulkDiscount requests.

This formal verification capability is essential when agents operate autonomously and can affect the real world. Organizations need mathematical certainty that their policies will behave as intended.

Unlike probabilistic AI models, enterprise security requires deterministic guarantees. Cedar policies always produce the same authorization decision for identical requests, regardless of evaluation order or system state. Cedar’s default deny, forbid wins, no ordering semantics help ensure predictable behavior.

// Policy evaluation order does not affect the authorization decision
permit(
    principal,
    action == AgentCore::Action::"ProcessRefund",
    resource
) when {
    context.input.refundAmount < 500
};

forbid(
    principal,
    action == AgentCore::Action::"ProcessRefund", 
    resource
) when {
    context.input.orderDate.offset(duration("90d")) < context.system.now
};

Whether the permit or forbid policy is evaluated first, a refund request over $500 will always be denied, and any refund issued more than 90 days after the order date will also be denied. This predictability gives enterprises confidence in their agent governance.

By choosing AgentCore Policy and Cedar, organizations can deploy autonomous agents with policies they can reason about mathematically, not only hope the agents work correctly. Cedar’s combination of expressiveness, readability, and formal verification means that you can design agents with the flexibility needed to function and the certainty security teams demand.

Automated reasoning has already proven its value across AWS, from AWS IAM Access Analyzer verifying access policies to provable security for network configurations. Applying these same techniques to agentic AI is a natural extension: as agents take on more responsibility, the need for mathematically grounded guarantees only grows. The neuro-symbolic approach we’ve described in this post—combining LLM flexibility with the rigor of automated reasoning—points toward a future where agents can be both more autonomous and more trustworthy, because the verification keeps pace with the autonomy.

Policy is now available as part of Amazon Bedrock AgentCore Gateway. To learn more about Cedar and its capabilities, visit the Cedar website, try the Cedar playground, or join the Cedar community on Slack.

For more information about Policy in Amazon Bedrock AgentCore Gateway, visit the AWS documentation or explore the AgentCore Gateway console.

If you have feedback about this post, submit comments in the Comments section below.