🚀 Redshift MCP Server

A plug-and-play Model Context Protocol (MCP) server for Amazon Redshift. Instantly connect LLMs and AI agents (Claude, ChatGPT, etc.) to your Redshift data — securely, safely, and with zero hassle.

🌟 Features

Production-ready: Secure, robust, and easy to deploy
MCP Standard: Exposes Redshift as standardized tools for LLMs/AI
Strict Access Modes: readonly, readwrite, admin (enforced by env var)
Auto-connect: Use environment variables or a .env file for seamless startup
No code changes needed: Just configure and run
Extensible: Add new tools or business logic easily
Works with Claude Desktop, ChatGPT, and any MCP client

⚡ Quickstart

Clone & Install

git clone <repo-url>
cd my-redshift-mcp
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Configure (recommended: create a .env file):

REDSHIFT_HOST=your-cluster.region.redshift.amazonaws.com
REDSHIFT_DATABASE=mydb
REDSHIFT_USER=myuser
REDSHIFT_PASSWORD=mypassword
REDSHIFT_PORT=5439
DB_MCP_MODE=readonly

Run the server
```
./scripts/run_server.sh
```
Connect from your LLM/MCP client (e.g., Claude Desktop)

🔗 What is Redshift MCP Server?

Redshift MCP Server is an open-source bridge between Amazon Redshift and modern AI assistants. It lets LLMs securely query, explore, and (optionally) modify your Redshift data using the Model Context Protocol (MCP). Perfect for:

Data teams enabling AI-powered analytics
Building AI workflows and automations
Secure, auditable LLM access to production data

About This Project

This repository provides a minimal, production-ready MCP server for Amazon Redshift using the FastMCP framework. It allows LLMs and AI agents to:

Query Redshift databases (with strict access controls)
List schemas and tables
Describe table structures
(Optionally) insert/update data, depending on configured mode

Key features:

Three access modes: readonly, readwrite, admin (see below)
Environment variable and .env support for easy configuration
Secure by default: starts in readonly mode
Compatible with Claude Desktop, ChatGPT, and any MCP client
Easy to extend with new tools or business logic

Features

This MCP server provides the following tools for Redshift database operations:

connect_db: Establish connection to a Redshift cluster (supports environment variables)
query: Execute SELECT queries with optional parameter binding
execute: Run INSERT, UPDATE, DELETE, or DDL statements
list_schemas: List all schemas in the database
list_tables: List tables in a specific schema
describe_table: Get detailed structure of a table
disconnect: Close the database connection

Auto-Connection Support

The server supports automatic connection on startup using environment variables. If all required environment variables are set, the server will connect automatically when it starts.

Access Modes and Query Safety

The server supports three access modes, controlled by the DB_MCP_MODE environment variable:

1. readonly (default)

Allows: SELECT, SHOW, DESCRIBE, etc.
Blocks: INSERT, UPDATE, DELETE, DROP, TRUNCATE, ALTER, CREATE, GRANT, REVOKE, COMMENT, SET, COPY, UNLOAD, VACUUM, ANALYZE, MERGE

2. readwrite

Allows: SELECT, INSERT, UPDATE, CREATE, etc.
Blocks: DELETE, DROP, TRUNCATE, ALTER, GRANT, REVOKE, COMMENT, SET, COPY, UNLOAD, VACUUM, ANALYZE, MERGE

3. admin

Allows everything (no restrictions)

If a forbidden statement is detected, the server will return a clear error message.

How to Set the Mode

Environment variable: DB_MCP_MODE=readonly (or readwrite, admin)

Claude Desktop config:

"env": {
  "DB_MCP_MODE": "readwrite",
  ...
}

Shell/local:

export DB_MCP_MODE=readonly
./run_server.sh

Installation

Clone or download this repository
Install dependencies:

pip install -r requirements.txt

Usage

Running the Server

Run the server directly:

python src/redshift_mcp_server.py

Or use the provided scripts:

./scripts/run_server.sh
# or on Windows:
scripts\run_server.bat

By default, the server runs using STDIO transport, which is suitable for integration with MCP clients like Claude Desktop.

Configuration for Claude Desktop

To use this server with Claude Desktop, add the following configuration to your Claude Desktop config file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%/Claude/claude_desktop_config.json

Option 1: With Environment Variables (Recommended)

{
  "mcpServers": {
    "redshift": {
      "command": "/Users/anshulpatre/Desktop/DB-mcp/my-redshift-mcp/venv/bin/python",
      "args": ["/Users/anshulpatre/Desktop/DB-mcp/my-redshift-mcp/src/redshift_mcp_server.py"],
      "env": {
        "REDSHIFT_HOST": "your-cluster.region.redshift.amazonaws.com",
        "REDSHIFT_DATABASE": "mydb",
        "REDSHIFT_USER": "myuser",
        "REDSHIFT_PASSWORD": "mypassword",
        "REDSHIFT_PORT": "5439",
        "DB_MCP_MODE": "readonly"
      }
    }
  }
}

With this configuration, the server will automatically connect to your Redshift database on startup. You won't need to use the connect_db tool manually.

Option 2: Without Environment Variables

{
  "mcpServers": {
    "redshift": {
      "command": "/Users/anshulpatre/Desktop/DB-mcp/my-redshift-mcp/venv/bin/python",
      "args": ["/Users/anshulpatre/Desktop/DB-mcp/my-redshift-mcp/src/redshift_mcp_server.py"]
    }
  }
}

With this configuration, you'll need to use the connect_db tool to establish a connection.

Environment Variables

The server supports the following environment variables:

REDSHIFT_HOST: Redshift cluster endpoint
REDSHIFT_DATABASE: Database name
REDSHIFT_USER: Username
REDSHIFT_PASSWORD: Password
REDSHIFT_PORT: Port number (default: 5439)
DB_MCP_MODE: Access mode (readonly, readwrite, admin)

When these environment variables are set, the server will:

Automatically connect on startup
Use them as defaults for the connect_db tool if no parameters are provided
Enforce query/statement restrictions based on the selected mode

Available Tools

1. connect_db

Connect to a Redshift database cluster. If environment variables are set, parameters are optional.

# With explicit parameters
{
  "tool": "connect_db",
  "arguments": {
    "host": "your-cluster.region.redshift.amazonaws.com",
    "database": "mydb",
    "user": "myuser",
    "password": "mypassword",
    "port": 5439
  }
}

# With environment variables set
{
  "tool": "connect_db",
  "arguments": {}
}

2. query

Execute SELECT queries to retrieve data.

{
  "tool": "query",
  "arguments": {
    "sql": "SELECT * FROM users WHERE created_at > %s",
    "params": ["2024-01-01"]
  }
}

3. execute

Execute data modification or DDL statements.

{
  "tool": "execute",
  "arguments": {
    "sql": "INSERT INTO users (name, email) VALUES (%s, %s)",
    "params": ["John Doe", "john@example.com"]
  }
}

4. list_schemas

List all user-created schemas.

{
  "tool": "list_schemas",
  "arguments": {}
}

5. list_tables

List tables in a specific schema.

{
  "tool": "list_tables",
  "arguments": {
    "schema": "public"
  }
}

6. describe_table

Get detailed information about a table's structure.

{
  "tool": "describe_table",
  "arguments": {
    "table": "users",
    "schema": "public"
  }
}

7. disconnect

Close the database connection.

{
  "tool": "disconnect",
  "arguments": {}
}

Security Considerations

Credentials: The recommended approach is to use environment variables in the MCP configuration file. This keeps credentials out of your code and tool calls.
Permissions: Use database users with minimal required permissions.
Query Safety: The server uses parameterized queries to prevent SQL injection. Query/statement type is checked and blocked according to the selected mode.
Network Security: Ensure your Redshift cluster is properly secured with VPC and security groups.

Development

To extend this server:

Add new tools using the @mcp.tool() decorator
Follow the FastMCP documentation for advanced features
Test thoroughly with your Redshift cluster

Troubleshooting

Connection Issues

Verify your Redshift cluster endpoint and credentials
Check network connectivity and security group settings
Ensure the cluster is active and accepting connections

Import Errors

If redshift-connector is not available, the server falls back to psycopg2
Install the appropriate connector based on your needs

License

MIT

Contributing

Feel free to submit issues or pull requests to improve this MCP server.

👤 Maintainer

This Redshift MCP Server is maintained and deployed by Anshul.