Lab 04

Lab 4: AgentCore Gateway Integration

Extend your mortgage assistant with external service capabilities using Amazon Bedrock AgentCore Gateway. Implement credit check functionality and learn how to securely integrate APIs, Lambda functions, and existing services as MCP-compatible tools.

What You’ll Build

  • Credit Check Service — real-time credit validation via Lambda
  • MCP Server Integration — convert services to Model Context Protocol tools
  • Secure Gateway Access — OAuth-protected service endpoints
  • Tool Composition — combine native and external tools seamlessly

Notebook

04-agentcore-gateway/mortgage_gateway.ipynb

What You’ll Learn

  • Service Integration Patterns — converting APIs to MCP-compatible tools
  • Gateway Configuration — setting up secure service endpoints
  • OAuth Implementation — protecting service access with authentication
  • Tool Composition — combining multiple tool types in agent workflows

Lab 05

Lab 5: AgentCore Identity

AgentCore Identity provides secure authentication, authorization, and credential management for AI agents. Enable agents to access external resources on behalf of users while maintaining strict security controls and audit trails.

Auth Types Supported

  • Inbound Auth — grant users access to invoke agents via AWS IAM or OAuth
  • Outbound Auth — grant agents access to AWS services and external resources
  • Workload Identities — specialized agent identities with fine-grained permissions
  • OAuth 2.0 Flows — 3-legged OAuth for user authorization flows

Notebook

05-agentcore-identity/InboundAuth/inbound_auth_runtime.ipynb

What You’ll Learn

  • Inbound vs. outbound authentication patterns
  • Configuring workload identities for agent-to-service auth
  • Implementing OAuth 2.0 flows for user authorization
  • Setting permission boundaries for agents acting on behalf of users