Redshift MCP Server
A Model Context Protocol (MCP) server that enables AI agents to interact with Amazon Redshift using natural language.
Features
- SQL Tools: Execute queries, list tables, describe schemas, get sample data.
- Resources: Connection status, table list.
- Sample Data: Pre-configured users, products, and orders tables.
- Architecture: See DESIGN.md for system diagrams.
🚀 Quick Start
For detailed instructions on Windows, macOS, and Linux, please refer to the SETUP.md guide.
Basic Steps (Local Testing):
- Start Postgres:
docker run -d -p 5432:5432 --name postgres -e POSTGRES_PASSWORD=password postgres - Install Deps:
pip install mcp redshift-connector pandas python-dotenv psycopg2-binary sqlalchemy - Seed Data:
python seed_redshift.py - Test:
python test_redshift_local.py
🔧 MCP Client Configuration
Add this to your MCP client configuration (e.g., Antigravity, Claude Code, or VS Code).
"redshift-mcp": {
"command": "py",
"args": ["redshift_mcp_server.py"],
"cwd": "c:/Users/deepa/OneDrive/Desktop/workspace/redis connection",
"env": {
"REDSHIFT_HOST": "localhost",
"REDSHIFT_PORT": "5432",
"REDSHIFT_DATABASE": "postgres",
"REDSHIFT_USER": "postgres",
"REDSHIFT_PASSWORD": "password"
}
}
☁️ AWS Deployment
See AWS_DEPLOYMENT.md for deployment options.