Optional ReadonlycreateOptional ReadonlycreateFlag controlling SSM parameter creation for construct resource references enabling
Optional ReadonlycrossReadonlydataAdmin roles granted key-admin access to the KMS key the module creates. Unused when kmsKeyArn
supplies an existing key.
Use cases: Administrative access, resource management, security control
AWS: IAM roles for Bedrock resource administration
Validation: Required; MdaaRoleRef[]
Optional ReadonlygatewaysBedrock AgentCore Gateway configurations (MCP servers), keyed by gateway name. Each gateway
references its tool targets by name from the sibling gatewayTargets map.
Use cases: exposing a unified MCP tool surface with per-tool authorization from the same module that owns the Lambdas it fronts
AWS: Amazon Bedrock AgentCore Gateway
Validation: Optional; NamedGatewayProps (map of gateway name to config)
Optional ReadonlygatewayBedrock AgentCore Gateway target definitions (MCP tool sources), keyed by target name. A gateway
attaches a target by naming its key in the gateway's targets list; each target may be
referenced by at most one gateway, and a target no gateway references is not deployed. A
knowledgeBase target generates a Lambda connector that exposes a knowledge base as a Retrieve
tool.
Use cases: defining MCP tool sources (Lambda tools, knowledge bases) once and referencing them from a gateway
AWS: AWS::BedrockAgentCore::GatewayTarget
Validation: Optional; NamedGatewayTargetProps (map of target name to config)
Optional ReadonlyguardrailNotification target and timing for the guardrail intervention alarms. Each guardrail
opts in with its own interventionAlarm thresholds; every alarm in the module
notifies the one target set here.
Use cases: routing guardrail intervention alarms to a security team, self-contained alerting with a module-created topic
AWS: CloudWatch Metrics Insights alarms + SNS notification
Validation: Optional; GuardrailAlarmsProps; required when any guardrail sets interventionAlarm
Optional ReadonlyguardrailsGuardrail configurations for AI safety, content filtering, and responsible AI deployment.
Use cases: AI safety controls, content filtering, responsible AI, content moderation
AWS: Bedrock Guardrails
Validation: Optional; NamedGuardrailProps (map of guardrail name to config)
Optional ReadonlyharnessesBedrock AgentCore Harness configurations - a declarative agent loop (model + system prompt +
tools) - keyed by harness name. A harness's guardrail.id and
tools[].agentCoreGateway.gatewayArn may use a config:<name> reference into the sibling
guardrails / gateways maps, resolved to the live resource.
Use cases: declarative tool-using/RAG agents
AWS: AWS::BedrockAgentCore::Harness
Validation: Optional; NamedHarnessProps (map of harness name to config)
Optional ReadonlykmsExisting KMS key ARN for encrypting all Bedrock resources in this module — knowledge bases, guardrails, the shared Lambda pool, and AgentCore gateways and harnesses all use this single key. If omitted, one customer-managed key is created automatically and shared across them.
When an existing key is provided, its key policy must already grant the required service and execution-role use (the module cannot mutate an imported key's policy).
Use cases: Customer-controlled encryption, security compliance, key reuse
AWS: KMS key for Bedrock resource encryption
Validation: Optional; String; must be valid KMS key ARN
Optional ReadonlyknowledgeKnowledge base configurations with S3/SharePoint data sources and custom parsing strategies. Enables document ingestion, embedding generation, and retrieval for RAG applications.
Use cases: Knowledge management, document processing, question-answering, RAG
AWS: Bedrock Knowledge Bases
Validation: Optional; NamedKnowledgeBaseProps (map of KB name to config)
Optional ReadonlylambdaLambda functions and layers for gateway tools, gateway interceptors, and knowledge base custom transformations. Enables custom business logic and external API integrations.
Use cases: Gateway tool logic, external integrations, function deployment
AWS: Lambda functions/layers
Validation: Optional; LambdaFunctionProps
ReadonlynamingMDAA naming implementation for consistent resource naming across all MDAA constructs
ReadonlyroleOptional ReadonlytagsOptional ReadonlyvectorVector store configurations for knowledge bases (OpenSearch Serverless or Aurora). Provides vector database storage for semantic search and RAG.
Use cases: Semantic search, RAG applications, knowledge retrieval, embedding storage
AWS: OpenSearch Serverless or Aurora vector stores
Validation: Optional; NamedVectorStoreProps (map of store name to config)
Optional ReadonlyvpcVPC endpoint sets, keyed by set name, giving harness sessions and knowledge base connectors a
private outbound path (no NAT/internet). A harness references one by name from its
networkConfiguration.vpcEndpoints, and a knowledgeBase gateway target from its
networkConfiguration.vpcEndpoints; every consumer referencing a set shares its endpoints.
Which endpoints exist is derived from the referencing consumers, so a set only states which VPC it serves, where created endpoints go, and how each derived endpoint is reached - created here, an existing one to wire to, or reached without an endpoint the set manages. Endpoint policies are derived and not configurable.
Use cases: private (no-NAT) harness sessions and knowledge base connectors, reusing centrally provisioned endpoints
AWS: AWS::EC2::VPCEndpoint (Interface and Gateway)
Validation: Optional; NamedVpcEndpointSetProps; each set must name a distinct VPC and be referenced by at least one harness or knowledge base connector
Flag controlling CloudFormation output and stack export creation for construct resources