MCP Server Snowflake

🚀 Snowflake MCP Server

A Model Context Protocol (MCP) server for performing read-only operations on Snowflake databases. This tool enables Claude to securely query Snowflake data without modifying any information.

🚀 Quick Start

The Snowflake MCP Server allows Claude to interact with Snowflake databases in a read - only manner. Here's how to get started:

Prerequisites

• Python 3.12+
• A Snowflake account with read - only permissions

Installation Steps

1. Copy the .env.example file and rename it to .env.
2. Configure the environment variables as follows:

   cp .env.example .env
   # Configure the following content in the .env file
   SNOWFLAKE_AUTH_TYPE=private_key  # or external_browser
   SNOWFLAKE_ACCOUNT_NAME=your_account_name
   SNOWFLAKE_USER_NAME=your_user_name
   SNOWFLAKE_PRIVATE_KEY_PATH=path_to_your_private_key  # if using private_key authentication
   

3. Install the dependencies:

   pip install -r requirements.txt
   

Running the Server

python mcp_server_snowflake.py

Integrating with Claude Desktop

1. Run the server in the terminal.
2. Open Claude Desktop and connect to the running MCP server to start interacting.

✨ Features

• Flexible Authentication: Connect to Snowflake in the following ways:
  • Authenticate using a service account and a private key.
  • Authenticate via an external browser for an interactive session.
• Connection Pool Management: With automatic background refresh to maintain long - running connections.
• Supported Operations: Query data, display database structure, describe query plans, explain execution plans, and limit result set sizes.
• Automatic LIMIT Clause: Automatically adds a LIMIT clause to prevent the return of large datasets.
• Input Validation: Uses Pydantic for input validation to prevent SQL injection.

💻 Usage Examples

Basic Usage

# Run the server
python mcp_server_snowflake.py

Advanced Usage

# Integrate with Claude Desktop
# Step 1: Run the server
python mcp_server_snowflake.py
# Step 2: Open Claude Desktop and connect to the running server

📚 Documentation

Configuration Options

Environment Variable Configuration

• SNOWFLAKE_CONN_REFRESH_HOURS: Set the automatic refresh interval for connections (default is 24).

  SNOWFLAKE_CONN_REFRESH_HOURS=4  # Set to 4 hours
  

Authentication Methods

Private Key Authentication

1. Create a user and generate a key pair in Snowflake. Refer to [Snowflake Documentation](https://docs.snowflake.com/en/user-guide/key - pair - auth) for specific operations.
2. Configure the environment variables:

   SNOWFLAKE_AUTH_TYPE=private_key
   SNOWFLAKE_PRIVATE_KEY_PATH=/path/to/your/private - key.pem
   

External Browser Authentication

1. Configure the environment variables:

   SNOWFLAKE_AUTH_TYPE=external_browser
   

2. After starting the server, the browser will automatically open for authentication.

Security Precautions

• Only allow read operations (SELECT, SHOW, DESCRIBE, EXPLAIN, and WITH statements).
• Input is strictly validated to prevent SQL injection attacks.
• Ensure the security of the .env file to avoid exposing sensitive information.
• Use HTTPS for communication to ensure data transmission security.

🔧 Technical Details

• Database Connection: Uses [Snowflake Connector Python](https://docs.snowflake.com/en/developer - guide/python - connector/python - connector) for database connection.
• Interaction with Claude: Implements interaction with Claude based on the [MCP Protocol](https://github.com/anthropics/anthropic - cookbook/tree/main/mcp).
• Data Validation: Uses Pydantic for data validation.
• Environment Variable Management: Uses [python - dotenv](https://github.com/theskumar/python - dotenv) to manage environment variables.

📄 License

No license information provided in the original document.

🛠️ Development Details

Static Type Checking

mypy mcp_server_snowflake/

Code Formatting

ruff format .

Running Tests

pytest