A production-grade, extensible Model Context Protocol (MCP) server for PostgreSQL, designed for LLM coding agents in VSCode, Cursor, and other IDEs.

Features

• MCP-compliant: Implements the MCP spec using @modelcontextprotocol/sdk
• PostgreSQL: Uses pg with connection pooling
• Modular Tools: All tools are organized in separate modules for extensibility
• Transports: Supports both HTTP (Streamable HTTP) and STDIO
• TypeScript: Fully typed, strict, and documented
• .env Config: Uses dotenv for configuration
• SSL Support: Configurable SSL connection to PostgreSQL
• Extensible: Easy to add custom tools and extend functionality
• Query Safety: Built-in protection against resource-intensive queries

Installation

npm install mcp-server-postgresql

Usage

import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { registerAllTools } from "mcp-server-postgresql/tools";

const server = new McpServer({
  name: "My MCP Server",
  version: "1.0.0"
});

// Register all built-in tools
registerAllTools(server);

// Start the server
server.start();

Extending with Custom Tools

You can create your own tools by implementing the ToolModule interface:

import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { ToolCallback } from "@modelcontextprotocol/sdk/server/mcp.js";
import { z } from "zod";

// Define your tool handler
const myCustomToolHandler: ToolCallback<{ param1: z.ZodString }> = async (args, extra) => {
  // Your tool implementation
  return { content: [{ type: "text", text: "Result" }] };
};

// Create a tool module
export function register(server: McpServer) {
  server.registerTool("my_custom_tool", {
    description: "Description of my custom tool",
    inputSchema: {
      param1: z.string()
    },
    outputSchema: {
      result: z.string()
    }
  }, myCustomToolHandler);
}

// Use in your server
import { register as registerCustomTool } from "./my-custom-tool.js";

const server = new McpServer({
  name: "My MCP Server",
  version: "1.0.0"
});

// Register built-in tools
registerAllTools(server);

// Register your custom tool
registerCustomTool(server);

server.start();

Built-in Tools

• list_tables: List all tables in a schema
• list_columns: List all columns for a table (with comments)
• find_related_tables: Show direct FK relationships for a table
• describe_relationship: Describe the relationship between two tables
• generate_erd_mermaid: Generate a Mermaid ERD diagram for a schema
• generate_erd_json: Generate a JSON graph of schema structure
• fuzzy_column_match: Fuzzy match a column by natural language phrase
• sample_column_data: Return sample data from a column
• run_query: Execute SELECT queries with safety measures:
  • Query complexity analysis
  • Rate limiting
  • Result size limiting
  • Query timeout protection
  • Structured responses

Configuration

Create a .env file in your project root:

# PostgreSQL Connection
PGHOST=localhost
PGPORT=5432
PGUSER=postgres
PGPASSWORD=yourpassword
PGDATABASE=mydb

# SSL Configuration (optional)
PGSSL=true                    # Enable SSL connection
PGSSL_REJECT_UNAUTHORIZED=true # Reject unauthorized SSL certificates

# Server Configuration
PORT=8080

Development

1. Clone the repo
2. Install dependencies
3. Build
4. Run

Development Scripts

• npm run build - Build the TypeScript project
• npm start - Run the built server
• npm run dev - Run the server in development mode using tsx
• npm run release:beta - Create and publish a beta release
• npm run release:patch - Create and publish a patch release
• npm run release:minor - Create and publish a minor release
• npm run release:major - Create and publish a major release

Dependencies

• @modelcontextprotocol/sdk: ^1.12.1
• dotenv: ^16.4.5
• pg: ^8.12.0
• string-similarity: ^4.0.4

License

MIT