BigQuery Studio Git repositories are GA for SQL scripts and notebooks

Google Cloud has made BigQuery Studio Git repositories generally available. The BigQuery release notes entry dated November 18, 2026 says analysts can manage and version control SQL scripts and notebooks with BigQuery Studio Git repositories, which Google describes as a folder-based integration with remote Git repositories, and labels the feature generally available (GA). The same capability had been listed as a preview since the release notes entry of May 18, 2026, with identical wording.
The practical change is that a remote Git repository can now be cloned directly into the BigQuery Studio file browser, edited in place, and committed and pushed from the console, under GA terms. For analytics teams that already keep dbt projects, Dataform code or shared SQL in GitHub or a similar host, it means saved queries and notebooks no longer have to live only inside the BigQuery project.
The documentation page for the feature, last updated on November 18, 2026, also sets clear limits. Git repositories sit in each user's own root folder, are intended for private use, and depend on Developer Connect for authentication. Teams planning to roll the feature out should read those limits before they replace any existing workflow.
··········
WHAT BIGQUERY STUDIO GIT REPOSITORIES DO.
According to Google's documentation, BigQuery Studio Git repositories let users version control SQL scripts, notebooks and other code assets from the file browser in BigQuery Studio. A connected repository appears as a folder under the user root node, and inside it a user can create SQL queries, notebooks, data canvases, data preparations, files and subdirectories. Any change made to a file is saved automatically, and files can be renamed, moved, copied, downloaded or deleted from the action menu.
Version control happens in the Source control pane, the tab that Google renamed from "Repository" to "Source control" in a separate change recorded on November 3, 2026. From that pane a user can open a line-by-line diff between the local version of a file and the last committed version, write a commit message, commit the change to the Git history of the local branch, and then push it to the remote repository. A Fetch from remote command brings in remote references and additional commits.
Branch handling is documented at a basic level. Users can check out a new local branch based on an existing local or remote tracking branch. Google states that a user cannot check out a new branch or switch branches while there are uncommitted changes, so work has to be committed first. The documentation pages reviewed for this report do not describe a pull or merge workflow or conflict resolution inside BigQuery Studio, so teams should plan to handle merges and pull requests on their Git host.
··········
BIGQUERY STUDIO GIT REPOSITORIES VS BIGQUERY REPOSITORIES.
The GA feature should not be confused with BigQuery repositories, an older capability that Google announced in preview in March 2025 and that the documentation still labels Preview. BigQuery repositories use workspaces, where each user edits a copy of the code before publishing changes to branches. Google's release note describes the new Git repositories as a folder-based integration with remote Git repositories, and the documentation positions them for each user's own version control.
........
Aspect | BigQuery Studio Git repositories | BigQuery repositories |
Launch stage | GA since November 18, 2026 | Preview |
Where code lives | Folder under the user root folder | Repository with workspaces |
Intended use | Private use by one user | Team collaboration through workspaces |
Authentication | Developer Connect account connector, or HTTPS or SSH | Connection to a third-party Git repository |
Version control | Diff, commit, push, fetch, branches in Source control | Commit and push from a workspace |
........
The sharing rule is the main difference for team leads. Google's documentation says Git repositories are restricted to the user root folder context and are intended for private use, and it tells users not to share these repositories with other users, even though they are technically created at the project level. Collaboration therefore happens through the remote Git host, where each analyst pushes branches and the team reviews them, rather than inside a shared BigQuery Studio folder.
··········
ROLES AND PERMISSIONS FOR BIGQUERY STUDIO GIT REPOSITORIES.
Before connecting a repository, an administrator has to enable the Developer Connect API in the Google Cloud project. Users then need the Developer Connect OAuth User role (roles/developerconnect.oauthUser) on the project, which Google says contains the permissions required to manage code with BigQuery Studio Git repositories. Custom roles or other predefined roles can also carry those permissions.
........
Role | What the documentation says |
Developer Connect OAuth User | Contains the permissions needed for Git repositories |
BigQuery User | Includes Developer Connect permissions |
BigQuery Studio User | Includes Developer Connect permissions |
BigQuery Admin | Includes token accessor permissions |
BigQuery Studio Admin | Includes token accessor permissions |
........
Two security notes in the documentation deserve attention from administrators. Users with the dataform.repositories.list permission, which the BigQuery Job User, BigQuery Studio User and BigQuery User roles include, can see all code assets in the Explorer panel of the project. Google also warns that any secrets shared with the Dataform service agent can potentially be accessed by users who can edit these assets. Projects where many people hold the BigQuery User role should review who can see which code before analysts start cloning repositories that contain sensitive logic.
··········
CONNECTING A REMOTE REPOSITORY TO BIGQUERY STUDIO.
A repository is created from the file browser, next to the user root node, by entering the URL of the remote Git repository. BigQuery Studio detects whether a Developer Connect account connector already exists and, if one is found, fills in the display name. If none exists, the user can create an account connector, and anyone who prefers not to use one can switch to HTTPS or SSH. Google notes that Git repositories use Developer Connect account connectors rather than Git repository connections.
Network access is a common point of failure. The documentation says the connection can fail if the remote repository is not open to the public internet, for example when it sits behind a firewall, and it points to the Dataform egress IP ranges for allow-listing. Organizations that enforce the dataform.restrictGitRemotes policy must first add the remote repository to the policy's allowed values.
Large repositories also need care. Google warns that repositories with many files, large files, many branches or a deep commit history take longer to clone, and the clone can exceed the operation timeout, which prevents the repository from being created. The Advanced options of the creation dialog offer a shallow clone with a configurable depth, and more commits can be fetched later.
··········
BIGQUERY STUDIO GIT REPOSITORY LIMITS FOR NOTEBOOKS AND FILES.
The documentation lists several limits that affect daily work. Notebook files stored in a Git repository mount cannot exceed 30 MB. When files change outside the mount, it can take up to 60 seconds for those changes to become visible inside the mount on the notebook runtime, and a cache setting adjusts that delay. File system operations within the mount location consume Dataform quota. Directories can only be moved within the same Git repository, and files can only be copied to directories within the same Git repository.
None of the BigQuery documentation pages reviewed on November 20, 2026 states a separate price for Git repositories. Queries and notebook runs started from files in a repository are billed like any other BigQuery work, so the usual cost controls still apply, as described in our report on BigQuery on-demand and editions pricing and in the piece on BigQuery project caps for slots and concurrency.
··········
SECURITY BULLETIN GCP-2026-084 AND THE GIT INTEGRATION.
One day after the GA entry, on November 19, 2026, the BigQuery release notes recorded a security item stating that a path traversal vulnerability in workspace directory handling and Git integration was discovered in Dataform, BigQuery Studio and Colab Enterprise, with a reference to security bulletin GCP-2026-084. The release note does not by itself say which configurations were affected or what customers need to do, so administrators should read the bulletin before enabling Git repositories widely and confirm whether any action is required in their projects.
··········
WHAT ANALYTICS TEAMS SHOULD CHECK BEFORE USING GIT IN BIGQUERY STUDIO.
The first check is access. The Developer Connect API must be enabled, users need the Developer Connect OAuth User role or a role that already includes its permissions, and the remote host must be reachable from Google's Dataform egress addresses. The second check is visibility, because broad BigQuery User grants let users list all code assets in a project, including repositories that hold sensitive business logic.
The third check is process. Because Git repositories in BigQuery Studio are private to each user and the documentation does not describe merges, review and integration remain on GitHub or the team's Git host. Analysts who write SQL with an assistant can commit the generated query and its revisions, which pairs well with the checks described in our guides on writing SQL with Gemini in BigQuery and on catching fan-out and date errors in AI-written SQL.
··········
BIGQUERY STUDIO GIT REPOSITORIES: CONFIRMED FACTS AND OPEN POINTS.
Google has confirmed in its release notes that BigQuery Studio Git repositories are generally available as of November 18, 2026, after a preview that began on May 18, 2026. The documentation confirms the folder-based model, the Source control pane with diff, commit, push, fetch and branch checkout, authentication through Developer Connect account connectors or HTTPS or SSH, and limits such as the 30 MB notebook size, the 60-second visibility delay inside mounts and Dataform quota consumption.
What depends on each organization is network reachability of the Git host, organization policies on allowed remotes, and how IAM roles are granted across the project. Google's pages do not list a separate price for the feature, do not document merge or conflict handling inside BigQuery Studio, and, for the November 19 security item, leave the details to bulletin GCP-2026-084. Teams that keep shared SQL and notebooks under review will still do that review on their Git host, with BigQuery Studio acting as the place where each analyst edits, commits and pushes.
··········
FOLLOW US FOR MORE.
·····
DATA STUDIOS
·····
datastudios.org




