BedrockTest

Terraform project to deploy an AWS Bedrock Agent with a connected Knowledge Base backed by S3 and OpenSearch.

The setup includes:

• Bedrock Agent

• Bedrock Knowledge Base

• S3 bucket (for documents)

• OpenSearch collection + index

• Configurable chunking strategies

Prerequisites

• Terraform ≥ 1.5.0

• AWS CLI configured (aws configure)

• Permissions to create:

  • S3 buckets

  • Bedrock resources

  • OpenSearch collections

  • IAM roles/policies

Setup

1. Clone the repo

git clone -b dev https://github.com/dmdcl/BedrockTest.git
cd BedrockTest

2. Initialize Terraform

terraform init

3. Create a terraform.tfvars file in the root folder with your values:

aws_region               = "us-east-1"
project_name             = "myproject"
environment              = "dev"
kb_s3_bucket_name        = "my-kb-bucket"   # Optional: if you want to bring your own bucket
agent_name               = "my-agent"
agent_model_id           = "anthropic.claude-v2"
agent_instruction        = "You are a helpful assistant."
agent_description        = "Agent for answering KB queries"
kb_name                  = "my-knowledge-base"
kb_description           = "KB with documents"
kb_embedding_model_id    = "amazon.titan-embed-text-v1"
opensearch_collection_name = "kb-collection"
opensearch_index_name    = "kb-index"
chunking_strategy        = "FIXED_SIZE"
fixed_size_max_tokens    = 300
fixed_size_overlap_percentage = 10
tags = {
  Owner = "me"
}

Modules

• modules/s3.tf → Creates the S3 bucket if you don’t provide your own.

• modules/main.tf → Core logic: agent, KB, and OpenSearch.

Configuration Notes

• By default, the repo creates its own S3 bucket for the KB.

• If you already have an S3 bucket you want to use:

  1. Pass its name via kb_s3_bucket_name in terraform.tfvars.

  2. Uncomment the corresponding lines in modules/main.tf:

  data "aws_s3_bucket" "kb" {
    bucket = var.aws_s3_bucket
  }

  And comment out the aws_s3_bucket.kb resource in modules/s3.tf.

Deploy

terraform plan
terraform apply

Confirm with yes. Terraform will create the agent, KB, S3 bucket (if enabled), and OpenSearch resources.

Destroy

To tear everything down:

terraform destroy

Upload Documents to KB

1. Copy your documents to the S3 bucket:

aws s3 cp ./documents s3://<kb-s3-bucket-name>/ --recursive

2. Start ingestion in Bedrock:

aws bedrock-agent start-ingestion-job \
  --knowledge-base-id <knowledge_base_id> \
  --data-source-id <data_source_id> \
  --region <aws_region>

You can get the values from the Terraform outputs:

terraform output

---

Test Your Agent

aws bedrock-agent-runtime invoke-agent \
  --agent-id <agent_id> \
  --session-id test-session \
  --input-text "Hello, what can you help me with?" \
  --region <aws_region>

---

Next Steps

• Monitor ingestion and queries via the AWS Console.

• Adjust chunking strategy, vector dimensions, and OpenSearch settings as needed.