mcp -starrocks
StarRocks MCP (Model Context Protocol) Server
claude mcp add --transport stdio starrocks-mcp-server-starrocks uv run --with mcp-server-starrocks mcp-server-starrocks \ --env STARROCKS_DB="default empty" \ --env STARROCKS_HOST="default localhost" \ --env STARROCKS_PORT="default 9030" \ --env STARROCKS_USER="default root" \ --env STARROCKS_PASSWORD="default empty"
How to use
The StarRocks MCP Server acts as a bridge between AI assistants and StarRocks databases. It supports direct SQL execution through read_query (SELECTs) and write_query, database exploration (listing databases and tables, and retrieving table schemas via starrocks:// resources), and provides detailed overviews of tables and databases (table_overview and db_overview) including column definitions, row counts, and sample data. It can also produce a Plotly chart directly from a query result via query_and_plotly_chart, and caches overviews in memory to speed up repeated requests with an option to bypass the cache. The server is configurable via environment variables and can be run in different transport modes; the recommended setup uses Streamable HTTP for RESTful interactions over JSON.
To use, ensure the MCP host points to the StarRocks MCP Server endpoint and configure connection details either via individual STARROCKS_* variables or a STARROCKS_URL connection string. The server exposes a JSON-based API over HTTP POST requests when using Streamable HTTP, enabling straightforward client integration without specialized SDKs.
How to install
Prerequisites:
- Access to a StarRocks FE endpoint (StarRocks server)
- The MCP runtime tool uv installed on your system (as provided in the README)
- Network connectivity between the MCP host and StarRocks instance
Installation steps:
- Ensure uv is available on your system. If needed, install it per your environment (refer to your environment's package manager or installation guide).
- Install or prepare the StarRocks MCP server package (mcp-server-starrocks) in your environment according to your usual workflow (the README sections demonstrate using uv to run the server with the appropriate environment).
- Start the MCP host with Streamable HTTP mode and configure the server via the provided mcp_config example:
uv run mcp-server-starrocks --mode streamable-http --port 8000
- Validate the connection by sending a test request to the Streamable HTTP endpoint (e.g., http://localhost:8000/mcp) using your MCP client tooling or a simple HTTP POST with a Test payload.
Configuration example (as JSON for the MCP host): { "mcpServers": { "mcp-server-starrocks": { "command": "uv", "args": ["run", "--with", "mcp-server-starrocks", "mcp-server-starrocks"], "env": { "STARROCKS_HOST": "default localhost", "STARROCKS_PORT": "default 9030", "STARROCKS_USER": "default root", "STARROCKS_PASSWORD": "default empty", "STARROCKS_DB": "default empty" } } } }
Additional notes
Environment variables:
- STARROCKS_HOST, STARROCKS_PORT, STARROCKS_USER, STARROCKS_PASSWORD, STARROCKS_DB can be used individually or via STARROCKS_URL for a single connection string. Examples of STARROCKS_URL: root:password@localhost:9030/dbname or starrocks://user:pass@host:9030/dbname.
- STARROCKS_OVERVIEW_LIMIT controls approximate total text size for overview generation (default 20000).
- STARROCKS_MYSQL_AUTH_PLUGIN can be set if your StarRocks deployment requires a specific authentication plugin.
- MCP_TRANSPORT_MODE selects the MCP transport (stdio, streamable-http, or sse; SSE is deprecated). Common issues: ensure the StarRocks endpoint is reachable from the MCP host, and confirm the port is accessible. If using STARROCKS_URL, make sure the URL is correctly formatted and that the provided credentials have enough permissions for the requested operations. The sse mode is deprecated; prefer streamable-http for new integrations.
Related MCP Servers
chunkhound
Local first codebase intelligence
skillz
An MCP server for loading skills (shim for non-claude clients).
mcp -odoo
A Model Context Protocol (MCP) server that enables AI assistants to securely interact with Odoo ERP systems through standardized resources and tools for data retrieval and manipulation.
agentql
Model Context Protocol server that integrates AgentQL's data extraction capabilities.
bzm
Official BlazeMeter MCP Server for AI-driven performance testing
mcp
Exasol MCP server. Provides knowledge about the Exasol database to an LLM through the Model Context Protocol.