How to Set Up a Snowflake MCP Server for Claude, Cursor & Codex

This post was written by an engineer at QueryPlane. QueryPlane is an app builder for your database: bring your own postgres db and you can create interactive applications to share with other developers, coworkers or even your customers. If you’re interested in trying it out, get started here.

---

MCP (Model Context Protocol) lets AI tools like Claude Desktop, Claude Code, Cursor, and Codex connect directly to your Snowflake data warehouse. Instead of copy-pasting schemas and query results into chat, the AI can inspect your tables, run queries, and leverage Snowflake’s Cortex AI features autonomously.

In this post, we’ll cover:

• What is MCP — How the protocol works
• Best Snowflake MCP servers — Official and multi-database options
• Setup guides — Step-by-step for Claude Desktop and Claude Code
• Security considerations — Keeping your data warehouse safe

What is MCP?

The Model Context Protocol is an open standard created by Anthropic that lets AI applications connect to external data sources and tools. Think of it as a USB-C port for AI — a standardized way for any AI client to talk to any data source.

For data warehouses like Snowflake, MCP servers expose tools for exploring schemas, running SQL, and accessing platform-specific features like Cortex AI — all callable by AI agents without manual context.

Best Snowflake MCP Servers

1. Official Snowflake Labs MCP

Snowflake-Labs/mcp (~275 stars) is maintained by Snowflake and provides deep integration with Snowflake’s AI features.

Features:

• Cortex Search (RAG over unstructured data)
• Cortex Analyst (structured data via semantic modeling)
• Cortex Agent (agentic orchestrator across structured and unstructured data)
• SQL execution and object management

The server requires a YAML config file (tools_config.yaml) to define which Cortex services and tools to enable. See the README for the full configuration reference.

Claude Desktop config:

{
  "mcpServers": {
    "snowflake": {
      "command": "uvx",
      "args": [
        "snowflake-labs-mcp",
        "--service-config-file", "/path/to/tools_config.yaml",
        "--connection-name", "default"
      ],
      "env": {
        "SNOWFLAKE_ACCOUNT": "your_account",
        "SNOWFLAKE_USER": "your_user"
      }
    }
  }
}

2. Google MCP Toolbox for Databases

googleapis/mcp-toolbox (~14,800 stars) supports Snowflake alongside PostgreSQL, MySQL, BigQuery, ClickHouse, MongoDB, and many more.

Features:

• Prebuilt generic tools (list_tables, execute_sql)
• Custom tools framework for domain-specific queries
• Connection pooling and IAM auth
• OpenTelemetry observability built in

Best for: Teams using multiple data sources, or anyone who prefers a generic MCP interface over Snowflake-specific features.

3. CentralMind Gateway

centralmind/gateway (~520 stars) supports Snowflake alongside PostgreSQL, MySQL, ClickHouse, and more. It auto-generates API endpoints from your database schema.

Which Snowflake MCP Server Should You Use?

Server	Stars	Official?	Cortex AI	Multi-database
Snowflake Labs MCP	~275	Yes	Yes	No
Google MCP Toolbox	~14,800	Google	No	Yes
CentralMind Gateway	~520	Community	No	Yes

For most developers, use the official Snowflake Labs MCP — it has the deepest Snowflake integration including Cortex Search and Cortex Analyst. If you need to connect to multiple data sources with one server, use Google MCP Toolbox.

Security Considerations

Connecting an AI agent to your Snowflake warehouse carries risk. Follow these practices:

• Use a dedicated role — create a role with minimal privileges for MCP access
• Grant SELECT only — restrict to read-only operations
• Use a dedicated warehouse — avoid contention with production workloads and control costs
• Limit schema access — grant access only to specific schemas and databases
• Don’t expose credentials in configs — use environment variables or key-pair authentication
• Enable query history — monitor what the AI runs via Snowflake’s QUERY_HISTORY

-- Create a read-only role for MCP
CREATE ROLE mcp_reader;
GRANT USAGE ON DATABASE mydb TO ROLE mcp_reader;
GRANT USAGE ON SCHEMA mydb.public TO ROLE mcp_reader;
GRANT SELECT ON ALL TABLES IN SCHEMA mydb.public TO ROLE mcp_reader;
GRANT USAGE ON WAREHOUSE mcp_warehouse TO ROLE mcp_reader;

CREATE USER mcp_user PASSWORD = 'secure_password' DEFAULT_ROLE = mcp_reader;
GRANT ROLE mcp_reader TO USER mcp_user;

Using MCP with QueryPlane

If you want the benefits of AI-powered database access without managing MCP servers, QueryPlane provides this out of the box. Connect your Snowflake warehouse and use AI to generate queries, explore schemas, and build dashboards — all from your browser with built-in access controls and team collaboration.

FAQ

What is a Snowflake MCP server?

A Snowflake MCP server is a process that implements the Model Context Protocol to let AI tools (Claude, Cursor, Codex, Windsurf) connect to your Snowflake data warehouse. It exposes tools for listing tables, describing schemas, running SQL, and accessing Cortex AI features.

How do I connect Claude to Snowflake?

Install the official Snowflake Labs MCP server, then add it to your Claude Desktop config or use claude mcp add in Claude Code. The AI can then query your warehouse directly.

Is it safe to connect AI to my Snowflake warehouse?

It can be safe with proper precautions: use a dedicated read-only role, a separate warehouse for MCP queries, limit schema access, and use key-pair authentication. Monitor queries via QUERY_HISTORY.

Can MCP access Snowflake Cortex features?

Yes. The official Snowflake Labs MCP server includes Cortex Search (RAG over unstructured data) and Cortex Analyst (structured data with semantic models). These let AI agents leverage Snowflake’s built-in AI capabilities.

Wrapping up

Snowflake’s official MCP server provides one of the deepest platform integrations in the ecosystem, with access to Cortex AI features alongside standard SQL execution. Start with a read-only role and a dedicated warehouse, then explore how AI-powered data access fits your analytics workflow.