Enables natural-language SQL interaction with Snowflake databases, allowing users to query tables, retrieve data, execute SQL commands, and manage database objects through conversational prompts.
Local-Snowflake-MCP-Server
A local Model Context Protocol (MCP) server that enables Claude Desktop to interact with a Snowflake database through natural-language SQL. Built in Python, this project demonstrates how modern LLMs can integrate with enterprise data systems using secure, local tooling.
🔗 Badges
Overview
This repository contains:
A Python MCP server (
snowflake_mcp_server.py)Instructions/Code in README for the claude_desktop_config.json file
Screenshots and documentation showing the MCP server running locally
Instructions for connecting Claude → MCP Server → Snowflake securely
This project demonstrates how to use MCP to build custom AI-powered database tooling.
Repository Structure
Getting Started
1. Clone the Repository
Install Dependencies
Set Up Environment Variables
Run the MCP Server
Configure Claude Desktop JSON
Example Claude Prompts “Run a query to list the tables in my schema.”
“Show the top 10 rows from MY_TABLE.”
“Explain what this SQL query does and suggest optimizations.”
“Create a table called DEMO_TEST with sample values.”
Security Notes (Important!) This repo does not include:
❌ Snowflake credentials ❌ Hardcoded passwords ❌ Personal account identifiers