AI Database Management: MCP Skills for Neon, Supabase & More

Setup Guide

The following instructions set up Neon MCP as the primary database skill, with Prisma MCP for migration management. This combination covers the complete schema lifecycle for a Next.js or Node.js project.

Step 1: Get Your Neon API Key

Log in to console.neon.tech, navigate to Account Settings › API Keys, and create a new key with read/write access to your projects. Store this key securely — you will reference it by environment variable name, never as a literal value in config files.

Step 2: Configure Neon MCP and Prisma Skill

// ~/.claude/settings.json
{
  "mcpServers": {
    "neon": {
      "command": "npx",
      "args": ["-y", "@neondatabase/mcp-server-neon"],
      "env": {
        "NEON_API_KEY": "your_neon_api_key_here"
      }
    },
    "prisma": {
      "command": "npx",
      "args": ["-y", "prisma-mcp-server"],
      "env": {
        "DATABASE_URL": "postgresql://user:pass@host/dbname"
      }
    }
  }
}

Step 3: Set Up Supabase MCP (Alternative)

For Supabase projects, replace the Neon configuration with Supabase MCP. Generate an access token at supabase.com/dashboard/account/tokens:

"supabase": {
  "command": "npx",
  "args": ["-y", "@supabase/mcp-server-supabase", "--access-token", "$SUPABASE_ACCESS_TOKEN"]
}

Step 4: Verify the Connection

After restarting your AI assistant, verify each skill with a simple query:

• "List my Neon projects" — should return your project names and IDs
• "Show the schema of the users table" — should return column definitions
• "Run EXPLAIN ANALYZE on SELECT * FROM users LIMIT 10" — should return a query plan

Use Cases

Database MCP skills are useful across the entire database lifecycle. Here are the four most impactful use cases with worked examples.

Schema Migrations

Ask your agent: "Add a stripe_customer_id column to the users table. It should be a varchar(255), nullable, with a unique index. Generate the Prisma migration and apply it to the development branch." The agent creates the Neon branch, generates the Prisma migration file, applies it, and reports success — all in a single conversation turn.

Database Branching per PR

Configure your CI pipeline to create a Neon branch for each pull request using Neon MCP. Each branch gets the same schema as main, migrations from the PR are applied to the branch, tests run against the isolated branch, and the branch is deleted when the PR closes. No migration conflicts, no shared development database state.

Query Optimization

"This query is taking 800ms on the subscriptions table. Here is the EXPLAIN ANALYZE output: [paste output]. What index should I add?" The agent reads the query plan, identifies the sequential scan on the user_id column, generates the CREATE INDEX CONCURRENTLY statement, and applies it — without taking the table offline.

Database Introspection and ORM Generation

"We have a legacy MySQL database. Introspect its schema and generate equivalent Prisma schema definitions for each table." Prisma MCP runs db pull, reads the existing schema, and outputs a complete schema.prisma file that your team can use as the starting point for a type-safe ORM layer.

Comparison Table

Use this table to choose the right database MCP skill based on your database provider and migration toolchain.

Frequently Asked Questions