Bedrock Knowledge Base
Vector index over a data source (S3, web, Confluence, SharePoint, Salesforce).
Configuration
Section titled “Configuration”| Setting | Type | Required | Default |
|---|---|---|---|
| Knowledge base name | Text | Yes | — |
| Description | Text | — | — |
| IAM role ARN | Text | Yes | — |
| Embedding model ARN | Text | Yes | — |
| Storage type Options: OpenSearch Serverless, Pinecone, Aurora PostgreSQL (pgvector), MongoDB Atlas | Choice | — | OPENSEARCH_SERVERLESS |
| Storage collection ARN | Text | — | — |
| Vector index name | Text | — | — |
| Field mapping (JSON) | Text | — | — |
| Tags | Key–value | — | — |
Connections
Section titled “Connections”| Socket | Direction | Accepts | Terraform arg |
|---|---|---|---|
| Data source | Input | any | — |
| Vector store (OSS / Pinecone / RDS) | Input | any | — |
| IAM role | Input | aws.iam-role | role_arn |
| Used by agent | Output | aws.bedrock-agent | — |