Govern AI agent access with Okta Cross App Access and Amazon Bedrock AgentCore – Part 2
In Part 1 you built one governed agent. Part 2 scales the pattern to an enterprise: Amazon Bedrock AgentCore Gateway becomes the single front door and redeems Okta's grant, Cedar policies bind each tool to the agent, a second agent runs on Amazon EKS, and AgentCore Observability traces every allow and deny beside Okta's log in Amazon CloudWatch.
First seen by Oriel on .
AI summary
Part 2 replaces direct agent-to-tool wiring with Amazon Bedrock AgentCore Gateway, Cedar policy enforcement, and centralized observability for governed Okta-delegated AI agent access at enterprise scale.
- The Gateway provides one front door, stores MCP target registrations, redeems each ID-JAG, and removes token-exchange and secret permissions from agent roles.
- Cedar policies use agent and scope claims with default deny, while MCP servers continue checking each user’s groups and recording authorization decisions.
- A reporting agent on Amazon EKS uses its own Okta identity, Gateway, credential provider, and narrowly scoped EKS Pod Identity role.
- AgentCore components emit OpenTelemetry spans, while Okta System Log events reach CloudWatch through Amazon EventBridge for joined audit queries in CloudWatch Omni.
- Targets need static tool schemas for referenced Cedar policies, clients must follow pagination, and the sample is conceptual, non-production code requiring security and legal review.
#ai agents #amazon bedrock agentcore #okta #cedar
Generated by gpt-5.6-sol. Check the original for details.
Aliases
| Source | External ID | URL |
|---|---|---|
| AWS Machine Learning Blog | 568f99493ce9fc7f0fbe42c57947610f9ca4666c | https://aws.amazon.com/blogs/machine-learning/govern-ai-agent-access-with-okta-cross-app-access-and-amazon-bedrock-agentcore-part-2/ Read discussion |