Install guide and config at curatedmcp.com

Supabase MCP Server: Query Your Database and Manage Auth with Natural Language

The official Supabase MCP server bridges the gap between AI agents and your PostgreSQL backend. Instead of manually writing database scripts or toggling between dashboards, you can now ask Claude, Cursor, or Windsurf to query tables, create schemas, manage Row Level Security policies, and deploy Edge Functions—all through conversation. It's the fastest way to let AI assistants interact with your Supabase project without exposing raw credentials to third parties.

What It Does

This server gives AI agents direct, sandboxed access to your Supabase project. Execute SQL queries, inspect database structure, generate TypeScript types from your schema, manage authentication users and providers, read/update storage buckets, and invoke Edge Functions—all without leaving your IDE or chat interface. You can ask Claude to "create a users table with email and role columns" or "deploy this Edge Function," and it happens instantly. The server handles RLS policy inspection, index performance checks, and environment variable management, making it practical for full-stack development workflows where your AI assistant needs to understand and modify your database layer in real time.

Maintained by the Supabase team, it stays current with platform updates and is trusted by thousands of developers building production applications.

How to Install

Run this command to install the latest version:

npx -y @supabase/supabase-mcp@latest

Add it to your Claude Desktop configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "supabase-mcp-server": {
      "command": "npx -y @supabase/supabase-mcp@latest",
      "env": {
        "SUPABASE_ACCESS_TOKEN": "your-access-token-here",
        "SUPABASE_PROJECT_ID": "your-project-reference-id"
      }
    }
  }
}

Generate an access token at app.supabase.com → Account → Access Tokens, and grab your project reference ID from your project settings.

Real-World Use Cases

• Schema generation & migrations: Ask Claude to inspect your current schema, then generate a migration script to add a new feature (e.g., "add a comments table with timestamps and user_id foreign key").
• RLS policy debugging: Paste an error log and ask your AI assistant to check your Row Level Security policies, identify the issue, and suggest fixes without leaving your editor.
• Edge Function iteration: Deploy a Postgres function or TypeScript Edge Function, test it with sample queries, and refine the logic—all in a single chat thread with Claude or Cursor.

Full install guides for Claude Desktop, Cursor, Windsurf, and more at CuratedMCP.