redshift

This Redshift MCP Server provides read-only access to an Amazon Redshift data warehouse via MCP-compatible clients. It exposes tools for querying, describing tables, and locating columns across schemas, enabling Cursor IDE and other MCP clients to fetch schema metadata, sample data, and statistics without compromising data integrity. The server runs over a Node.js-based TypeScript implementation and reads its Redshift connection details from the DATABASE_URL environment variable. To use the server, configure it in a project-wide or global mcp.json file, then connect your MCP client. The available tools—query, describe_table, and find_column—let you execute safe read-only queries, inspect table structures, and search for tables containing particular column names.

Once connected, Cursor can automatically discover the redshift-mcp server and present tools under Available Tools. Clients can request resources such as Schema Listings, Table Schemas, Sample Data, and Statistics, which Cursor can render for exploration and reasoning. The server enforces read-only transactions and input sanitization, and redacts sensitive fields in sample data to reduce exposure of PII.

Typical usage flows include querying for table lists, inspecting a table’s schema, and locating tables with specific columns. Examples of tool interactions include: asking the query tool to show all tables in a schema, using describe_table to reveal a table’s columns and data types, and invoking find_column to locate tables containing a certain column pattern.

Prerequisites:

• Node.js 16 or higher
• TypeScript tooling (tsc) and build steps configured in package.json
• Access to an Amazon Redshift cluster
• Cursor IDE installed (for optimal experience)

Installation steps:

# 1. Clone the repository or download the source
git clone https://github.com/yourorg/redshift-mcp-server.git
cd redshift-mcp-server

# 2. Install dependencies
npm install

# 3. Build the TypeScript code (produces dist/index.js)
npm run build

Configuration:

# Set the Redshift connection URL as an environment variable when running
export DATABASE_URL="redshift://username:password@hostname:port/database?ssl=true"

# Start the server in production mode
npm start

# Or run the development version with live reload (if configured)
DATABASE_URL="redshift://username:password@hostname:port/database?ssl=true" npm run dev

Optional: if you are integrating with MCP clients using stdio transport, use a transport manifest like the example in the README and ensure node dist/index.js or the built entrypoint is accessible.

Tips and considerations:

• Ensure your Redshift cluster allows connections from your running environment and that the DATABASE_URL uses ssl=true for secure connections.
• The server performs read-only queries to prevent data modifications; always verify your tooling is using read-only capabilities.
• Sensitive data in sample results is redacted automatically; never rely on sample data for production secrets.
• If you change the Resource or Tool schemas, update the server’s handlers (ListResourcesRequestSchema, ReadResourceRequestSchema, ListToolsRequestSchema, CallToolRequestSchema) accordingly.
• When deploying, consider environment-specific DATABASE_URL values (dev/stage/prod) and avoid embedding credentials in code or config files that are not securely stored.