How to Enable Direct Supabase Schema Editing (No Migration Files) Across IDEs (Bolt, Windsurf, Curso
Hi Supabase community,
We’re using Supabase as the backend for an Expo app, and we’re designing a workflow that starts in Bolt but expands across multiple IDEs and AI-assisted environments — including Windsurf, Cursor, Claude, and Codex — all connected to the same Supabase project.
Here’s our situation and what we’d like clarification on:
1. Starting point and workflow
We typically begin projects in Bolt, where AI agents can directly create and edit Supabase tables and schema without generating migration files.
As projects scale, we connect them to GitHub, allowing team members to continue development in their preferred IDEs while keeping Supabase connected to the same backend.
2. Core question
We want to replicate Bolt’s ability to make direct schema and data changes to Supabase (e.g., creating or editing tables, columns, relationships, and policies) — without using migration files — across any IDE or AI editor in our workflow.
Specifically:
How can we configure Supabase (CLI, API, or permissions) so developers or AI tools in other IDEs can directly apply schema changes to the database (e.g., via SQL or Supabase CLI) without creating migration files?
Is using supabase db query or a similar direct SQL workflow the correct and supported approach for this purpose?
Are there recommended safeguards or project configurations to make this workflow secure and maintainable?
3. Team structure and synchronization
Bolt remains our central hub, but we want Windsurf, Cursor, and other IDEs to work in tandem, all capable of directly interacting with the same Supabase backend.
Our source control is GitHub, so we’ll later sync or generate migration files (via supabase db pull or supabase db diff) for version control once schema changes stabilize.
4. Goal
We’d like to establish a clear, supported process where:
Schema changes can be applied live to Supabase (no migrations needed during active development).
3 Replies
Later, these direct changes can be pulled into migration files for CI/CD and long-term version tracking.
AI editors (Claude, Codex, etc.) can safely and automatically perform these updates in the background — just as Bolt does natively.
Could you please confirm the recommended approach or best practice for enabling this kind of direct-edit workflow in Supabase?
Thank you for your help — we’re trying to make this the foundation of our multi-IDE, AI-assisted development workflow.
Best regards
If you want multiple Ai editors to interact with your database then you can use the supabase mcp for most (if not all) of the main ones https://supabase.com/docs/guides/getting-started/mcp. Then you can create migration files by pulling the remote schema https://supabase.com/docs/reference/cli/supabase-db-pull and checking these into github.
CLI Reference | Supabase Docs
CLI reference for the Supabase CLI
So far when it comes to using AI tools with the database there is not really a standard/recommended process and using one AI tool alone tends to cause a large amount of difficulty maintaining migrations let alone several. I think the standard it still the use of migration files.