MCP Server: AWS S3 Knowledge Base Retrieval

What This Repository Does

@modelcontextprotocol/server-aws-kb-retrieval is an MCP server that connects build-cli to AWS S3, enabling direct file access and exploration within a session. It uses the AWS SDK under the hood and picks up credentials from the standard AWS credential chain (environment variables, ~/.aws/credentials, or IAM role).

Common use cases:

Browse S3 bucket structure and sample files during pipeline development
Read a Parquet or CSV file from S3 to understand schema and content
Cross-reference AWS Glue job configuration while writing or reviewing pipeline code

How to Use It

Add the server to your Claude Code MCP configuration (~/.claude/settings.json):

{
  "mcpServers": {
    "aws-s3": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-aws-kb-retrieval"],
      "env": {
        "AWS_ACCESS_KEY_ID": "your-key",
        "AWS_SECRET_ACCESS_KEY": "your-secret",
        "AWS_REGION": "eu-west-1"
      }
    }
  }
}

Prefer IAM role-based credentials (for example, via AWS SSO) over long-lived access keys. Once configured, restart build-cli and the server becomes available automatically.

Security considerations:

Use a least-privilege IAM role — read-only S3 access is sufficient for most development tasks.
Do not connect to buckets containing PII or regulated data without confirming compliance with your data steward and security team.
Never store long-lived AWS credentials in the config file — use environment variables or AWS SSO.

Relevance to the Hub

This server is the primary way DMDQP software engineers inspect S3-resident data (Parquet, CSV, Glue configs) directly inside a build-cli session, removing the need to download files or switch to a separate S3 browser. It is particularly useful during data pipeline development and review when understanding the actual data shape is critical.