Snowflake MCP Server

Snowflake MCP Server 🎿

A Model Context Protocol (MCP) server that enables Claude to interact with your Snowflake data warehouse through natural language.

🌟 What is this?

Instead of logging into Snowflake UI and writing SQL manually, simply ask Claude in natural language and let it query your data warehouse for you!

This MCP server gives Claude the ability to:

• ✅ Execute SQL queries on Snowflake
• ✅ List tables in your schemas
• ✅ Describe table structures
• ✅ Check data freshness (when tables were last updated)

How It's Different

📋 Table of Contents

• Quick Start
• Installation
• Configuration
• Available Tools
• Usage Examples
• How It Works
• Security
• Extending
• Troubleshooting
• FAQ
• Contributing
• License

🚀 Quick Start

See QUICKSTART.md for a 5-minute setup guide!

# 1. Clone the repo
git clone https://github.com/Legolasan/snowflake_mcp.git
cd snowflake_mcp

# 2. Install dependencies
pip install -r requirements.txt

# 3. Configure credentials
cp .env.example .env
nano .env  # Add your Snowflake credentials

# 4. Test connection
python test_connection.py

# 5. Add to Claude Code (see Configuration section)

📦 Installation

Prerequisites

• Python 3.8 or higher
• Snowflake account with access credentials
• Claude Code CLI (for MCP integration) - Get it here

Note: You do NOT need a separate Claude API key! The MCP server runs locally and only requires Snowflake credentials. See FAQ for details.

Install Dependencies

pip install -r requirements.txt

This installs:

• mcp - Model Context Protocol SDK
• snowflake-connector-python - Snowflake database driver
• python-dotenv - Environment variable management

⚙️ Configuration

1. Snowflake Credentials

Copy the example environment file:

cp .env.example .env

Edit .env with your Snowflake credentials:

SNOWFLAKE_USER=your_username
SNOWFLAKE_PASSWORD=your_password
SNOWFLAKE_ACCOUNT=abc12345.us-east-1
SNOWFLAKE_WAREHOUSE=COMPUTE_WH
SNOWFLAKE_DATABASE=PROD_DB
SNOWFLAKE_SCHEMA=PUBLIC

Finding your account identifier:

• Your Snowflake URL: https://abc12345.us-east-1.snowflakecomputing.com
• Your account ID: abc12345.us-east-1

⚠️ Security: Never commit .env to git! It's protected by .gitignore.

2. Claude Code MCP Configuration

Add this to your Claude Code MCP settings:

File: ~/.config/claude-code/mcp.json

Option A: Using .env file (Recommended)

{
  "mcpServers": {
    "snowflake": {
      "command": "bash",
      "args": [
        "-c",
        "source /path/to/snowflake_mcp/.env && python /path/to/snowflake_mcp/server.py"
      ]
    }
  }
}

Option B: Direct environment variables

{
  "mcpServers": {
    "snowflake": {
      "command": "python",
      "args": ["/path/to/snowflake_mcp/server.py"],
      "env": {
        "SNOWFLAKE_USER": "your_username",
        "SNOWFLAKE_PASSWORD": "your_password",
        "SNOWFLAKE_ACCOUNT": "your_account",
        "SNOWFLAKE_WAREHOUSE": "COMPUTE_WH",
        "SNOWFLAKE_DATABASE": "your_database",
        "SNOWFLAKE_SCHEMA": "PUBLIC"
      }
    }
  }
}

3. Restart Claude Code

After updating the MCP configuration, restart Claude Code to load the server.

🛠️ Available Tools

1. `query_snowflake`

Execute any SQL query on Snowflake and get formatted results.

Parameters:

• sql (required): SQL query to execute
• limit (optional): Max rows to return (default: 100)

Example prompts:

• "Show me the top 10 customers by revenue"
• "How many orders were placed yesterday?"
• "What's the average order value this month?"

2. `list_tables`

List all tables in a schema with row counts.

Parameters:

• schema (optional): Schema name (defaults to configured schema)

Example prompts:

• "What tables are available?"
• "List all tables in the ANALYTICS schema"

3. `describe_table`

Get detailed structure of a table (columns, types, constraints).

Parameters: