Runs SQL analytics on SageMaker Catalog asset metadata tables exported as Apache Iceberg in S3 Tables. Covers governance queries, asset growth tracking, ownership audits, time-travel over catalog state, and metadata quality analysis. Applies when querying catalog inventory, finding assets without descriptions, comparing catalog snapshots, or auditing data ownership. Trigger phrases: catalog inventory SQL, how many assets, assets without descriptions, asset growth over time, who owns this data, c
This skill lets an agent run SQL analytics over Amazon SageMaker Catalog asset metadata that has been exported as a daily-snapshot Apache Iceberg table in the AWS-managed S3 Tables bucket. It solves the problem of understanding and governing a large data catalog without building custom ETL: instead of scraping the catalog UI, users query catalog inventory, count assets, find governance gaps, audit ownership, and compare historical snapshots directly in SQL.
The skill provides a decision tree for when to use catalog analytics versus other tools, plus concrete AWS CLI recipes for the full lifecycle: checking whether the data export is configured (`aws datazone get-data-export-configuration`), enabling it with optional KMS encryption, verifying Lake Formation SELECT/DESCRIBE grants, and querying the `asset_metadata.asset` table. It documents key columns (snapshot_time partition key, asset_id, resource_type_enum, business_description, extended_metadata, timestamps) and ships ready-made queries for current state counts, assets lacking descriptions, 30-day growth trends, time-travel comparisons across snapshots, and ownership rollups. The table is read-only and queries must always filter on snapshot_time to avoid scanning all history.
It targets data governance teams, data platform engineers, and analysts operating SageMaker Unified Studio who need to audit catalog quality, track asset growth, and answer ownership questions. It works best with the AWS MCP server for sandboxed execution and audit logging, but runs anywhere AWS credentials are configured.
It queries the SageMaker Catalog asset metadata exported as a read-only Apache Iceberg table in the AWS-managed aws-sagemaker-catalog S3 Tables bucket, partitioned by snapshot_time and refreshed once daily.
A configured SageMaker Unified Studio (DataZone) domain with data export enabled, the S3 Tables federated catalog registered in Glue (s3tablescatalog), Lake Formation SELECT and DESCRIBE grants on the table, and AWS CLI credentials. First data appears within 24 hours of enabling export.
The table stores every daily snapshot; without a snapshot_time filter a query scans all historical snapshots and returns duplicate rows. The skill defaults to DATE(snapshot_time) = CURRENT_DATE for current state.
No. The table is read-only. Adding descriptions, tags, or glossary terms is handled by other tools (Glue Discovery put-form-type / associate-glossary-terms), not this skill.
No, it works in any environment with configured AWS credentials, but it works best with the AWS MCP server which adds sandboxed execution and audit logging.
Quick Setup:
.claude/skills/Repository
aws/agent-toolkit-for-aws