SingleStore MCP Server – AI Database Operations

SingleStore MCP Server connects MCP-compatible AI clients to SingleStore Helios services for database, workspace, notebook, file, scheduled job, and usage workflows. Agents can use it to inspect accessible organizations and workspaces, run SQL against connected workspaces, and automate selected SingleStore management tasks through natural language.

#database#sql#helios

Overview

SingleStore MCP Server is the official Model Context Protocol server for SingleStore Helios and SingleStore database services. It lets MCP-compatible clients such as Claude Desktop, Claude Code, Cursor, Visual Studio Code, Windsurf, Gemini CLI, LM Studio, Goose, and Qodo Gen interact with SingleStore through a local MCP server. The server is useful when an AI assistant needs live context from a SingleStore environment rather than relying on static documentation or copied query results.

What the MCP server enables

The server exposes tools, resources, and prompt templates for database management, notebook workflows, job scheduling, and usage tracking. Documented capabilities include retrieving user and organization information, selecting an organization, listing workspace groups and workspaces, retrieving deployment regions, running SQL queries on connected workspaces, and working with Shared Workspaces. It can also manage notebooks, list sample notebooks, create notebooks in a user's personal workspace, manage files in personal and shared spaces, create scheduled jobs that run notebooks, and inspect historical job executions and status.

When to use it

Use this MCP server when an AI agent needs to explore a SingleStore Helios account, answer operational questions about databases and workspaces, or help developers build with SingleStore. Practical examples include asking an assistant to inspect accessible workspaces before querying data, generate SQL against a connected workspace, create a notebook in a personal space, create a scheduled notebook job, or help scaffold a Drizzle ORM project using SingleStore context.

Connection and authentication

The standard local setup uses the uvx command with singlestore-mcp-server start over stdio. SingleStore also documents initialization commands such as uvx singlestore-mcp-server init and client-specific variants for supported clients. In the standard local configuration, no API keys, tokens, or environment variables are required because authentication is handled through browser-based OAuth when the server starts. Docker is also documented as a local stdio-style launch path, but Docker requires MCP_API_KEY because browser-based OAuth is not supported inside locally hosted containers.

Key considerations

Users need a SingleStore Helios account and API access credentials. The server can expose powerful actions, including SQL execution, workspace information access, notebook creation, file management, and job scheduling, so the connected SingleStore account and API key should be scoped and protected appropriately. Docker deployments should avoid hard-coding secrets in client configuration; reference MCP_API_KEY from a secure local secret source. Only use this server with MCP clients and user permissions that are appropriate for the target SingleStore environment.

Frequently Asked Questions

When should an AI agent use the SingleStore MCP Server?

Use it when an AI agent needs live access to SingleStore Helios context, such as exploring organizations and workspaces, running SQL against a connected workspace, creating notebooks, managing files, or scheduling notebook jobs from an MCP-compatible client.

What does the SingleStore MCP Server add to an AI agent's capabilities?

It gives the agent MCP tools, resources, and prompt templates for SingleStore database services, including workspace and database management, notebook workflows, scheduled jobs, usage tracking, and structured SQL query execution.

What can the AI agent access or manage through this server?

Documented tools cover user and organization information, organization selection, workspace groups, workspaces, deployment regions, SQL queries, Shared Workspaces, notebooks, files, scheduled jobs, and job execution history, subject to the authenticated user's SingleStore permissions.

How is authentication configured for the SingleStore MCP Server?

The standard `uvx singlestore-mcp-server start` local configuration does not require API keys or environment variables because the server starts a browser-based OAuth flow. Docker-based operation requires an API key exposed as `MCP_API_KEY`, represented in this file through the local `SINGLESTORE_MCP_API_KEY` secret reference.

Which transport should be used for the SingleStore MCP Server?

Use the standard stdio transport with `uvx` for normal local MCP clients because it supports browser-based OAuth. Use the Docker stdio launch only when a containerized setup is required, and provide an API key because OAuth is not supported for locally hosted containers.