If you use Lovable to build applications, you know how fast it can consume build credits. Once those credits run out, upgrading to a higher plan can get expensive.
You do not have to stop working or pay for more Lovable credits. If you already have a personal subscription to Claude or Codex, you can use their desktop development tools to write code locally and sync it back to your Lovable project.
In this step-by-step guide we show you how to set up your environment so you can keep building for free.
Required Tools
To work on your project outside of the Lovable browser interface, you will need a few tools on your computer:
GitHub Desktop
Visual Studio Code (VS Code)
AI Extensions: the Codex extension (for ChatGPT users) or Claude Code (for Anthropic users) inside VS Code.
Step 1: Create a Developer Context File
Before moving your project to your local computer, you need to ensure that any outside AI agent understands exactly how your app is structured.
Open your project in Lovable.
Turn on Plan Mode.
Send the following prompt to the Lovable chat:
"Write an 'about this app' MD file for the next developer that tells them everything they need to know about this application, its dependencies, the systems it uses, APIs it integrates with, and anything else that they would need to know to consistently continue to develop this application."
Read through and approve the plan. Lovable will generate a file named about-this-app.md in your codebase. This file acts as the foundational guide for outside AI models.
Step 2: Connect Lovable to GitHub and Clone Locally
Now export your code out of Lovable so your local code editor can access it.
In the Lovable editor, open the dropdown menu and go to Settings.
Connect your account to GitHub and click the button to create/sync the repository.
Open GitHub Desktop on your computer.
Select Add and Clone Repository, find your new Lovable project repo, and clone it to your local hard drive.
Open the cloned folder inside Visual Studio Code.
Step 3: Configure Local AI Instructions
To ensure your local AI subscription reads the context file you generated in Step 1, you need to create strict instruction files.
In the root directory of your project inside VS Code, create a new file depending on which AI tool you use:
Paste the following instruction into your file:
"Always review the About This App MD before starting work. Update that MD file as a working document if you make changes that would impact the truthfulness of that document."
Because these files exist in the project, your local extensions should read them and learn the application’s structure before writing any new code.
Step 4: Develop Features and Sync Changes
Now you can prompt your local AI tool just like you would inside Lovable. For example, you can open your local terminal or AI chat extension and ask it to write a new feature or modify an existing piece of text.
Once the local AI tool finishes writing or editing the files on your computer:
Open GitHub Desktop.
You will see a list of the modified files. Type a summary of your changes (e.g., "updated menu text").
Click Commit to main and then click Push origin to sync the changes back to GitHub.
Lovable constantly watches your GitHub repository. As soon as the push is complete, Lovable will automatically ingest the new code, rebuild the preview, and update your project app in the cloud without deducting any build credits.
Bonus Content: Handling Database Migrations
When you build features locally that require changes to your database schema, your local AI tools cannot run those migrations directly on Lovable's servers.
Pushing the code back to Lovable, you might see that the frontend UI is updated but the data isn't saving correctly because the database table doesn't exist yet. You have two options to fix this:
The Daily Credit Method: Lovable gives users 5 free renewal credits every day. You can simply wait until your daily refresh, open the Lovable editor, and type "Apply database migrations" to let Lovable finalize the database setup.
The Database URL Method: If you cannot wait, you can use a migration toolkit script to extract your Lovable project's direct database connection URL. You can feed this URL directly into your local Codex or Claude Code terminal, allowing your local AI to execute the SQL migration script remotely without using Lovable's infrastructure.
Opichi offers a Lovable Migration Toolkit to help with this process.
For more details and a complete tutorial, follow along with our YouTube video Out of Lovable Credits? Do this Instead:
https://youtu.be/xdbcFUOUj2o?si=XDtsQSGXWCHeqWbi