Quick answer
Git records tracked source changes and supports diffs, branches, and rollback of code. It does not automatically protect databases, uploaded media, credentials, or files excluded from the repository. Pair a Git workflow with an appropriate website backup and a verified deployment process.
Before you start
Use a repository you are authorized to access and a local project folder with a clean baseline. Decide which files belong in source control and keep secrets, database dumps, large backups, and generated caches excluded.
What to do
- Create a branch or isolated worktree for the task and record the starting commit. Keep unrelated work separate.
- Ask the assistant to implement the narrow change and show the diff. Review file additions and deletions as well as modified lines.
- Run the checks appropriate to the project, then approve a specific commit or build artifact for deployment.
- Back up current live files and data as needed, deploy the approved version, and record the destination and readback result.
Check the result
Confirm that the deployed files correspond to the approved source or generated artifact. Test the public page and business interaction after release.
If something does not work
If live edits happened outside Git, reconcile them before deploying an older local copy. Never force a database or customer-data rollback merely because the code branch was reverted.
Related help
- Create a recoverable checkpoint before an AI website deployment
- Review an AI-generated website change before deployment
- Choose a selective rollback instead of restoring everything
Official references
Reviewed October 31, 2026. Interfaces, vendor availability, and account features can change. Check your current account before making changes.
