What it does

The StarRocks MCP Server acts as a bridge between AI assistants and StarRocks databases. It enables direct SQL query execution (both read and write operations), database and table exploration, schema inspection, system metrics access, and data visualization through Plotly charts. The server supports multiple transport modes (stdio, HTTP, Streamable HTTP) and caches database and table metadata in memory to optimize repeated requests.

Who it's for

Data engineers and analysts integrating StarRocks with Claude or other AI assistants for ad-hoc querying, schema discovery, and data-driven analysis without requiring client-side database tools.

Common use cases

• Execute SELECT queries directly on StarRocks databases from Claude conversations
• Explore database schemas and retrieve table metadata, including column definitions and row counts
• Generate visualizations from query results using Plotly
• Access internal StarRocks metrics and system state via the proc:// resource
• Build in-context AI workflows that combine data retrieval with analysis

Setup pitfalls

• Connection credentials (STARROCKS_PASSWORD, connection URL) must match your StarRocks cluster; misconfigured credentials will fail with database-level errors
• The process requires write permissions to the local filesystem for metadata caching; ensure the cache directory is writable
• Network connectivity to the StarRocks FE service (default port 9030) is required; verify STARROCKS_HOST is reachable from the MCP server process
• SSE (Server-Sent Events) transport mode is deprecated; use Streamable HTTP or stdio for new deployments