SupaForge

What SupaForge MCP server does

SupaForge MCP server brings Supabase project drift management into an MCP-compatible agent workflow. It can inspect differences between environments and help apply changes so a target reflects a source. The comparison extends beyond database tables and columns to project configuration and access controls, including RLS policies, policy coverage, storage buckets, storage policies, authentication settings, Edge Functions, scheduled jobs, webhooks, Realtime configuration, Vault, extensions, roles, grants, reference data, and migration history.

This makes the server relevant when database migrations alone do not describe the complete state of a Supabase project. It can be used with Supabase Cloud, self-hosted Supabase deployments, and local clones.

How it works

The SupaForge MCP server uses the project’s comparison and synchronization capabilities through the MCP interface. Scans identify differences between a source and target environment. The command-line implementation can return machine-readable JSON, which is useful when an agent or automation pipeline needs to interpret results.

Write operations preview their intended changes first. Applying a fix requires an explicit apply action, while operations that could remove data or broaden access require the destructive-change option. Fixes are executed in dependency order and run in a transaction, with rollback if an operation fails.

A proof mode can rehearse a migration against a temporary copy of the target. SupaForge applies the change only when the rehearsed result matches the source, providing an additional check before modifying the real environment. The underlying schema and data checks use DBDiff, which is installed automatically as a native binary.

Setup and configuration

Install the CLI package globally with:

npm install -g @akal/supaforge

The repository also documents a Homebrew installation and a GitHub Action. Initialize a project configuration with supaforge init, then use the MCP entry point through supaforge mcp. The README identifies integrations for Cursor, VS Code, and Claude, including one-click installation options. Environment-specific setup, self-hosted configuration, and connection details are covered in the project’s full CLI documentation, but are not specified in the supplied material.

Tools and capabilities

The SupaForge MCP server supports agent workflows for:

• Comparing Supabase environments for drift.
• Inspecting differences in database structure, policies, permissions, and platform configuration.
• Applying synchronized changes after review.
• Using proof-before-change checks for migrations.
• Producing machine-readable results for automation.
• Working with hosted, self-hosted, and local project environments.

The wider CLI also supports snapshots, restoring a snapshot into an empty database, and cloning a project with its data. These capabilities are available through SupaForge’s command-line interface; the supplied material does not define separate MCP tool names.

Limitations and notes

SupaForge MCP server is designed for Supabase environment comparison and synchronization, not for general-purpose database administration. Changes that may lose data or open access are not applied automatically and require the explicit destructive-change option. The MCP server is included in the repository as a standalone build and is used through the supaforge mcp command rather than as a separately published package. The project is licensed under MIT.