Fine-grained access control for Amazon Bedrock Managed Knowledge Base tools
Learn how to expose an Amazon Bedrock Managed Knowledge Base to AI agents as MCP tools through Amazon Bedrock AgentCore Gateway while keeping per-user access control. A request interceptor injects verified identity, Policy in AgentCore validates it independently, and a resource-based policy restricts direct access to the knowledge base.
First seen by Oriel on .
AI summary
This AWS post presents a layered design for enforcing per-user document access when Amazon Bedrock agents retrieve data through MCP tools and a shared Managed Knowledge Base.
- Managed Knowledge Base ACL awareness filters Retrieve and AgenticRetrieveStream results using S3 document ACL metadata and the supplied userContext email.
- An AgentCore Gateway interceptor overwrites userContext with the email from a validated Amazon Cognito JWT, preventing clients or LLMs from choosing another identity.
- Policy in AgentCore checks the modified identity, while a resource-based policy restricts knowledge-base searches to the gateway execution role.
- ACL-enabled documents without matching permissions are silently excluded; ACL changes apply after the next Amazon S3 synchronization, and the knowledge base does not authenticate users itself.
- The sample deploys a shared agent, knowledge base, Cognito authentication, gateway, web app, and AgentCore runtime with one AWS CDK stack.
#amazon bedrock #access control #mcp #cognito
Generated by gpt-5.6-sol. Check the original for details.
Aliases
| Source | External ID | URL |
|---|---|---|
| AWS Machine Learning Blog | 2144423e5e4caeacf910447d8e0923290056c400 | https://aws.amazon.com/blogs/machine-learning/fine-grained-access-control-for-amazon-bedrock-managed-knowledge-base-tools/ Read discussion |