Supabase MCP Server

This is a Model Context Protocol (MCP) server that connects to Supabase and allows you to query the feature_suggestions table.

Prerequisites

• Node.js (v16 or higher)
• npm
• Supabase project with credentials

Setup

1. Make sure your .env file contains the following Supabase credentials:
   SUPABASE_URL=your_supabase_url SUPABASE_ANON_KEY=your_supabase_anon_key
2. Install the required dependencies:
   npm install

Running the Server

Run the MCP server using:

Or use the npm script:

Connecting to AI Tools

Cursor

1. Open Cursor and navigate to Cursor Settings.
2. Under the Features tab, tap + Add new MCP server under the MCP Servers section.
3. Enter the following details:
   • Name: Supabase
   • Type: command
   • Command: npx tsx /path/to/mcp-server.ts
4. You should see a green active status after the server is successfully connected.

Claude Desktop

1. Open Claude desktop and navigate to Settings.
2. Under the Developer tab, tap Edit Config to open the configuration file.
3. Add the following configuration:
   { "mcpServers": { "supabase": { "command": "npx", "args": ["tsx", "/path/to/mcp-server.ts"] } } }
4. Save the configuration file and restart Claude desktop.

Available Tools

query_feature_suggestions

Query the feature_suggestions table in your Supabase database.

Parameters:

• limit (number, optional): Maximum number of records to return (default: 100)

Example usage in AI tool:

Or with a limit:

Troubleshooting

• If you encounter connection issues, make sure your Supabase credentials are correct.
• Check the console output for any error messages.
• Ensure that the feature_suggestions table exists in your Supabase database.