@squadbase/mcp-redshift

MCP (Model Context Protocol) server for Amazon Redshift.

Features

• Execute SQL queries against Redshift clusters
• Browse databases, schemas, and tables
• Inspect table schemas with distribution and sort key info
• Get table statistics and storage information
• Designed for AI/LLM data analysis workflows

Installation

# No installation needed - use npx
npx -y @squadbase/mcp-redshift

# Or install globally
npm install -g @squadbase/mcp-redshift
mcp-redshift

Usage

Direct Execution

npx -y @squadbase/mcp-redshift

Via Launcher

npx -y @squadbase/mcp redshift

MCP Client Configuration

Add to your mcp.json:

Via Launcher (recommended):

{
  "mcpServers": {
    "redshift": {
      "command": "npx",
      "args": ["-y", "@squadbase/mcp", "redshift"],
      "env": {
        "REDSHIFT_HOST": "your-cluster.region.redshift.amazonaws.com",
        "REDSHIFT_PORT": "5439",
        "REDSHIFT_DATABASE": "dev",
        "REDSHIFT_USERNAME": "admin",
        "REDSHIFT_PASSWORD": "your-password"
      }
    }
  }
}

Direct:

{
  "mcpServers": {
    "redshift": {
      "command": "npx",
      "args": ["-y", "@squadbase/mcp-redshift"],
      "env": {
        "REDSHIFT_HOST": "your-cluster.region.redshift.amazonaws.com",
        "REDSHIFT_PORT": "5439",
        "REDSHIFT_DATABASE": "dev",
        "REDSHIFT_USERNAME": "admin",
        "REDSHIFT_PASSWORD": "your-password"
      }
    }
  }
}

Environment Variables

| Variable | Description | Required | |----------|-------------|----------| | REDSHIFT_HOST | Cluster endpoint | Yes | | REDSHIFT_PORT | Port number (default: 5439) | Yes | | REDSHIFT_DATABASE | Database name | Yes | | REDSHIFT_USERNAME | Username | Yes | | REDSHIFT_PASSWORD | Password | Yes |

Available Tools

ping

Health check - returns server name and version.

Input: None

Output:

{
  "ok": true,
  "server": "mcp-redshift",
  "version": "0.0.1"
}

redshift_info

Get connection requirements and configuration info.

Input: None

Output:

{
  "ok": true,
  "message": "TODO: implement Redshift connectivity via host/username/password",
  "required_env": [
    "REDSHIFT_HOST",
    "REDSHIFT_PORT",
    "REDSHIFT_DATABASE",
    "REDSHIFT_USERNAME",
    "REDSHIFT_PASSWORD"
  ]
}

execute_sql

Execute a SQL query against Redshift.

Input:

{
  "query": "SELECT * FROM public.users LIMIT 10",
  "database": "dev",
  "max_rows": 1000
}

list_databases

List all databases in the Redshift cluster.

Input: None

list_schemas

List all schemas in a database.

Input:

{
  "database": "dev"
}

list_tables

List all tables in a schema.

Input:

{
  "schema": "public",
  "include_views": false
}

describe_table

Get detailed information about a table including columns, distribution style, and sort keys.

Input:

{
  "schema": "public",
  "table": "users"
}

Output includes:

• Column names, types, nullability
• Distribution key (distkey) columns
• Sort key columns
• Table storage info (diststyle, sortkey type, size)

sample_table

Get sample rows from a table for data exploration.

Input:

{
  "schema": "public",
  "table": "users",
  "limit": 10
}

get_table_stats

Get statistics and storage information for a table.

Input:

{
  "schema": "public",
  "table": "users"
}

Output includes:

• Row count
• Size in MB
• Unsorted percentage
• Statistics accuracy
• Last analyzed timestamp

Current Status

This package provides stub implementations. The tools return placeholder data to demonstrate the API structure. Full implementation with actual Redshift connectivity via pg driver is planned.

License

MIT