AWS KB Retrieval

Overview

The AWS Knowledge Base Retrieval MCP server enables your AI agent to perform semantic search over Amazon Bedrock Knowledge Bases. This allows your AI to answer questions using your organization's private documents — PDFs, internal wikis, S3 files, and more — using retrieval-augmented generation (RAG).

This is ideal for enterprises that have ingested private documentation into Bedrock and want their developers' AI assistants to have access to that proprietary knowledge while coding.

Setting Up Amazon Bedrock Knowledge Base

1. Log in to the AWS Console and navigate to Amazon Bedrock → Knowledge Bases.
2. Click Create Knowledge Base and configure a data source (S3 bucket with your documents).
3. Select an embedding model (e.g., amazon.titan-embed-text-v1).
4. After creation, note the Knowledge Base ID (format: XXXXXXXXXX).
5. Configure an IAM user or role with the following permissions:
   • bedrock:RetrieveAndGenerate
   • bedrock:Retrieve
   • bedrock:ListKnowledgeBases

Using AWS SSO / Profiles

Instead of hardcoding credentials, you can use AWS credential profiles:

jsonCOPY
{
  "env": {
    "AWS_PROFILE": "my-bedrock-profile",
    "AWS_REGION": "us-east-1",
    "BEDROCK_KB_ID": "your-knowledge-base-id"
  }
}

Run aws configure --profile my-bedrock-profile first to set up the profile.