SchemaFlow

You connect your AI IDE to SchemaFlow’s MCP server to access live PostgreSQL and Supabase schemas. This enables real-time schema context to improve code generation and database-aware actions inside your editor or IDE plugins.

How to use

You enable real-time schema access by configuring your MCP-enabled IDE client with the provided MCP servers. Each server points to SchemaFlow’s MCP endpoint and uses a secure token to access your cached schema data. Once connected, your AI assistant can query tables, columns, relationships, and other schema components in real time, helping you generate accurate queries, models, and code.

Common usage patterns include asking your AI to show you database schemas, list tables or relationships, inspect functions or indexes, or perform a broader schema analysis. Your IDE will leverage the MCP connection to fetch schema metadata on demand without exposing or transferring actual data.

Tip: Keep your MCP token secure. Tokens are unique to your account and revokable. If you revoke or rotate a token, update the MCP configuration in your IDE to continue accessing cached schema data.

How to install

Prerequisites: None are required on the client side to consume MCP data if you are using the hosted SchemaFlow MCP endpoints. If you want to verify token-based access or test locally, you’ll typically use the configuration snippets below in your IDE.

Step 1: Obtain an MCP token from the SchemaFlow dashboard after connecting your database and loading its schema. Step 2: Add the MCP server configuration to your IDE as shown in the examples. Step 3: Open your IDE and confirm the MCP connection is active, then try a sample query like asking for the database schema or table list.

If you prefer to copy a ready-to-use configuration, paste one of the following MCP server blocks into your IDE settings to enable the connection.

Configuration examples

{
  "mcpServers": [
    {
      "name": "schemaflow_cursor",
      "type": "http",
      "url": "https://api.schemaflow.dev/mcp/?token=your-token-here",
      "args": []
    },
    {
      "name": "schemaflow_windsurf",
      "type": "http",
      "url": "https://api.schemaflow.dev/mcp/?token=your-token-here",
      "args": []
    },
    {
      "name": "schemaflow_vscode",
      "type": "http",
      "url": "https://api.schemaflow.dev/mcp/?token=your-token-here",
      "args": []
    }
  ]
}

Security notes

Only schema metadata is accessed and cached on the client side. No actual data from your database is retrieved through MCP. Tokens are revokable and connections are encrypted to protect your schema information.

Notes and tips

If you manage multiple IDEs, you can configure each with its own named MCP server entry to keep environments isolated and easy to manage.

Available tools

get_schema

Retrieves complete database schema information including tables, columns, relationships, functions, triggers, enums, and indexes from PostgreSQL or Supabase databases.

analyze_database

Performs comprehensive database analysis including performance insights, security assessment, and structural recommendations for PostgreSQL and Supabase databases.

check_schema_alignment

Validates your PostgreSQL or Supabase schema against best practices and identifies potential issues with actionable recommendations.