Lovable Just Got Its MCP - Here's What That Unlocks

One of the things I've been most excited about from Lovable lately isn't a new UI feature or a template library. It's the MCP - and it just launched today.

If you're not familiar, MCP (Model Context Protocol) is a way for AI tools to connect directly to services. Think of it like a bridge - instead of copying and pasting between tabs, your AI assistant can talk directly to Lovable. Create projects, send messages to the agent, read files, deploy, connect integrations, check analytics - all from one place.

I've been sharing this with the Lovable community and honestly, I'm thrilled. This is going to open doors for a lot of people. I've already started experimenting - I updated my migration skill for moving from Lovable Cloud to your own Supabase, and I'm hoping to build the reverse too (Supabase to Lovable Cloud). For anyone who needs either direction, this is going to make life much easier.

And here's the thing - it's not just for Claude Code. You can connect the Lovable MCP in Claude on the web (claude.ai), and it's going to work with Claude Coworker too. If you use Cursor, Codex, or any other tool that supports MCPs, you can connect it there as well. The protocol is the same.

That's exciting because it means teams can use this too. If you're sharing your Lovable projects with a bigger or more technical team, they can now access project data at whatever level they need. Before, it wasn't easy to give a technical team member visibility into Lovable Cloud. Now it's much more accessible.

What You Can Do With the Lovable MCP

Here's the practical breakdown. Whether you're building your first app or managing multiple projects, these tools give you new ways to work:

What	How	Who it's for
Create and manage projects	create_project, list_projects, get_project	Everyone
Send prompts to the Lovable agent	send_message - same as typing in the editor	Everyone
Read project files	list_files, read_file - browse your entire codebase	Developers, teams
Query your database	query_database - SQL access to Lovable Cloud	Developers, migrations
Deploy projects	deploy_project	Everyone
Manage integrations	add_connector, list_connectors	Everyone
Set persistent instructions	set_project_knowledge, set_workspace_knowledge	Teams, power users
View analytics	get_project_analytics, get_project_analytics_trend	Everyone
Remix projects	remix_project	Everyone

For people who are just getting started with Lovable, the most useful ones are create_project, send_message, and deploy_project - you can build and deploy without ever opening the browser. For more technical users and teams, the database access and file reading tools are where it gets really powerful.

The Game Changer: Database Access

Now, the part that genuinely surprised me.

If you've used Lovable Cloud - that's Lovable's managed database - you know the deal: it works great, but you don't get a Supabase dashboard or direct SQL access. That's been one of the most common questions I get in the community: "how do I see my data?"

With the Lovable MCP, you can now run SQL queries directly against your Lovable Cloud database. Full access. Inspect your schema, browse your data, check your RLS policies. And here's the big one - you can even read auth.users, including the encrypted password hashes.

Why does that matter? Because it means you can migrate your entire database to your own Supabase project - including user credentials. Your users keep their exact same login. No password reset emails. No "please create a new account." That was previously considered impossible.

I tested this with a full production migration and it works. I'll walk through the entire process in my next post.

What This Means for Lovable Cloud Users

To be clear - Lovable Cloud is a solid managed database. It works well for a lot of projects and I'm not here to tell you to migrate. But if you've been feeling limited, or if you need things like custom auth providers, email templates, or a staging environment - now you have a clear path to get there without losing anything.

You now have:

• Full SQL access to inspect your schema and data
• The ability to export everything (including auth users with credentials)
• A path to migrate to your own Supabase if you need more control
• Better visibility for your team into what's in the database

How to Set It Up

It takes about two minutes:

1. Open Claude Code (or any MCP-compatible tool)
2. Type /mcp
3. Select Lovable
4. Authenticate with your Lovable account

That's it. Once connected, you can start working with your projects immediately.

For the full docs: Lovable MCP documentation

Important Note

The Lovable MCP is still in early stages. Tools and behavior may change as the team iterates on it. What I'm sharing here is what works as of May 2026 - and I'll keep updating as things evolve.

What's Next

In my next post, I'll walk through the complete migration process - taking a Lovable Cloud project and moving it entirely to your own Supabase. Database, auth users (with original passwords), storage, edge functions, frontend code. 40 steps, 2 manual clicks, everything else automated.

I've tested this in Claude Code, and I'm planning to test it in Codex and Cursor too - and any other tool that can connect an MCP. These things excite me. We needed this and it's bigger than people realize. Don't be afraid to connect your tools - whether it's Claude, Cursor, ChatGPT, or whatever comes next. The more connected your workflow, the more you can do.

I'm publishing the full migration guide at the same time as this post. If you're ready to migrate, head over to How to Migrate from Lovable Cloud to Your Own Supabase for the step-by-step walkthrough.