Branching, committing, and opening pull requests from inside the Databricks workspace

Databricks Repos syncs a folder in your workspace with a remote Git repository, letting you manage branches, commits, and pull requests without leaving the Databricks UI. This supports standard Git workflows so notebooks and code can be version-controlled alongside CI/CD pipelines.

1 · Learn the must-know

• 1You must connect a Git provider (e.g., GitHub, GitLab, Bitbucket, Azure DevOps) via personal access token in User Settings before cloning a repo into Databricks Repos.
• 2The branch selector in the Repos UI lets you create a new branch from an existing one and switch branches directly, checking out the latest commit for that branch.
• 3Changes to notebooks and files in a repo are tracked as uncommitted changes; you use the built-in dialog to stage, commit with a message, and push to the remote.
• 4Databricks Repos does not support merging branches directly in the UI; merges and pull requests are completed on the Git provider's site, though Databricks provides a link/button to open the PR creation page.
• 5Pulling changes from the remote updates the workspace copy, and conflicts must be resolved in the Git provider (or locally), since Databricks Repos has no built-in merge conflict resolution tool.
• 6Notebooks are stored in source format (e.g., .py, .sql, .ipynb) in Repos so diffs are readable in Git, unlike the default Databricks notebook (.dbc) format.

2 · Check your understanding