This project provides a Snowflake MCP Server, enabling integration of Snowflake's advanced capabilities—including Cortex AI, object management, and SQL orchestration—into the Model Context Protocol (MCP) ecosystem. It targets developers and data professionals seeking to leverage Snowflake data and AI features through MCP-compatible clients like Claude Desktop or Cursor, offering a unified interface for complex data operations and AI-driven insights.

How It Works

The MCP server acts as a bridge, translating MCP client requests into Snowflake operations. It dynamically creates MCP tools based on a YAML configuration file, enabling features like Cortex Search for RAG, Cortex Analyst for semantic data modeling, Cortex Agent for orchestrating data retrieval, and direct SQL execution. The server supports various Snowflake authentication methods via the Snowflake Python Connector and offers multiple transport mechanisms (stdio, SSE, streamable-HTTP) for flexible deployment.

Quick Start & Requirements

• Primary Install/Run: Execute via uvx snowflake-labs-mcp --service-config-file <path_to_config.yaml>.
• Prerequisites: Snowflake account, Snowflake Python Connector, uvx package. Requires Snowflake credentials (username/password, key pair, OAuth, SSO, MFA).
• Configuration: A services/configuration.yaml file drives tool creation and enables/disables service groups (Cortex, Object Management, Query Execution, Semantic Views).
• Deployment: Supports local execution, container deployment (Docker, Docker Compose), and various transport protocols.
• Docs: Snowflake Python Connector documentation for authentication. MCP Introduction for ecosystem context.

Highlighted Details

• Cortex AI Integration: Provides tools for Cortex Search (RAG), Cortex Analyst (semantic modeling), and Cortex Agent (orchestration).
• Data & Object Management: Enables basic Snowflake object operations (create, drop, describe, list) and secure SQL execution with configurable permissions.
• Semantic View Querying: Facilitates discovery and querying of Snowflake Semantic Views, including metrics and dimensions.
• Flexible Authentication & Transport: Supports all Snowflake Python Connector authentication methods and multiple transport protocols (stdio, SSE, streamable-HTTP).

Maintenance & Community

Bug reports and feedback should be submitted via GitHub issues. Specific details on maintainers, sponsorships, or community channels (like Discord/Slack) are not provided in the README.

Licensing & Compatibility

The README does not specify a software license. This omission represents a significant adoption blocker, as license type and compatibility for commercial use or closed-source linking cannot be determined.

Limitations & Caveats

The server is designed to be used with an MCP Client; it is not a standalone application. Object management tools are basic, with Snowsight recommended for advanced operations. Only Cortex Agent objects pre-configured in Snowflake are supported as tools. Deprecation notices exist for older CLI arguments (--account-identifier, --pat). Startup time can take several seconds due to Snowflake connection and configuration validation.