awslabs/mcp

Overview

This skill provides tools and patterns for building with Amazon Aurora DSQL. It helps manage schemas, run queries, and perform safe migrations while enforcing DSQL-specific constraints like async indexes and tenant isolation. Use it to develop scalable, distributed applications that require IAM-based authentication and MCP tool integration.

How this skill works

The skill exposes MCP tools to run readonly SELECT queries, execute transactional DDL/DML, and inspect table schemas. It enforces DSQL rules such as one-DDL-per-transaction, CREATE INDEX ASYNC, and batching updates under 3,000 rows. Reference documents (development guide, language patterns, examples, troubleshooting) should be consulted before schema changes or implementation.

When to use it

• Creating or evolving schemas for multi-tenant applications
• Performing safe, staged data migrations in Aurora DSQL
• Running ad-hoc read queries via MCP tools with tenant scoping
• Implementing application-layer referential integrity patterns
• Setting up or managing DSQL clusters and IAM-based connections

Best practices

• Always read the development guide before schema changes
• Execute each DDL in its own transact call; run async indexes separately
• Include and validate tenant_id in all queries and schema designs
• Batch updates under 3,000 rows and avoid ALTER TABLE with DEFAULTs
• Serialize arrays/JSON as TEXT and prefer connection pooling in prod

Example use cases

• Create a tenant-scoped table, add async indexes, and verify with get_schema
• Add a column safely: ALTER TABLE, batch UPDATE existing rows, then index
• Validate parent existence with readonly_query before inserting child rows
• List tables using readonly_query on information_schema and inspect schema
• Connect via psql with IAM token using provided cluster scripts for debugging