MCP tools for Amazon Aurora DSQL: Query execution and schema management

This article explains how to set up and use the Amazon Aurora DSQL MCP server to integrate database operations into AI coding assistants like Kiro, Claude Code, and Cursor.

• MCP server exposes seven tools: readonly_query, transact, get_schema, dsql_search_documentation, dsql_read_documentation, dsql_recommend, and dsql_lint
• Purpose-built for Aurora DSQL with IAM authentication, distributed transaction awareness, and PostgreSQL compatibility guidance
• Setup requires Python 3.10+, uv package manager, AWS credentials, and an Aurora DSQL cluster with appropriate IAM permissions
• Read-only by default; enable --allow-writes flag only when schema changes or data modifications are needed
• Supports schema-aware query generation, schema evolution with guardrails, documentation-grounded development, and query performance analysis
• Runs locally on your machine with TLS encryption; no remote proxy or cloud component between IDE and cluster
• Configure multiple MCP instances for active-active multi-Region clusters to verify replication and compare data across Regions

The MCP server streamlines database workflows by keeping database interactions within your development environment while maintaining security through IAM authentication and write protection.