Confirm your e-mail, then open Dashboard → Bitbucket.
2
Connect Bitbucket
Create a token with the backup and restore scopes
CYBBACK reads Bitbucket with one token that holds both the read scopes (backup) and the write scopes (restore). Use an account that administers the workspace.
Option A - API token with scopes (any plan)
Atlassian account settings → Security → Create API token with scopes → app Bitbucket.
Restore scopes (write does not include read - keep the read scopes): write:repository:bitbucket, admin:repository:bitbucket, write:pullrequest:bitbucket, admin:pipeline:bitbucket, write:webhook:bitbucket.
Copy the token; you will also enter the e-mail of the Atlassian account (not the Bitbucket username).
Option B - Workspace access token (Premium plan, recommended)
Not tied to a person, and eligible for Bitbucket's higher (scaled) rate limits on Standard/Premium plans with 100+ users. One workspace per token.
Rate limits
With an API token Bitbucket allows 1,000 requests per hour on repository data. CYBBACK paces itself and waits on 429 answers, so the first backup of a large workspace can take several hours; the next ones only read what changed.
All repositories (default) of the configured workspaces, or Selected repositories (click Load repositories).
Content: code, pull requests, pipelines, deployment environments, branch restrictions (with branching model and default reviewers), webhooks, permissions and deploy keys, optional build statuses.
Incremental forever
A repository's git bundle is rebuilt only when one of its branches or tags moved. Pull requests are read from the last watermark; the configuration is captured at every run with a frozen copy, so an older backup restores the settings of its own date.
Out of scope
Bitbucket issues and wikis were removed by Atlassian on 2026-08-20 and are not backed up.
4
Bring Your Own Bucket
Use your own S3-compatible storage
Want full data sovereignty? Point CYBBACK at your S3-compatible bucket - AWS S3, GCS, Scaleway, OVH, Wasabi, MinIO. Your data, your provider, your region.
Provision the bucket
In your S3 provider, create a private bucket (no public access, versioning recommended).
Create an access key / secret key with permissions limited to that bucket.
Required IAM actions:PutObject, GetObject, DeleteObject, ListBucket.
Resilience: CYBBACK stores an encrypted snapshot of S3 credentials with each backup. Old backups remain restorable even after you switch providers.
Disk usage: a typical mirror of 100 active repos (~2 GB combined) takes ~500 MB on S3 thanks to git's natural compression.
5
First backup
Run your first backup
Back on the Dashboard tab, click Start backup. A daily schedule (05:00 Europe/Paris) is created automatically after the first backup.
Follow the progress live in the history or on the Events page.
A repository that fails (permissions, empty repository…) is listed with its reason under the backup - the others are still saved.
6
Restore
Restore - same or another workspace
Selection. Tick a repository to restore it entirely, or open it (workspace → project → repository) and choose item by item, with « Select all (N) » on every list: branches and tags one by one (under their name, a safe restored/<date>/ prefix or a name you type), files and folders as they were at the backup date (one commit on a branch, with an optional pull request), open pull requests, branch restrictions and merge checks, default reviewers, webhooks, deploy keys, pipeline variables and schedules, deployment environments and variables, permissions, repository settings, branching model and Downloads files. « Compare with Bitbucket » shows the branches deleted or rewritten since the backup and pre-selects them.
Files and folders. In the Source tab, the Restore files or folders card shows the file tree of any branch or tag as it was on the backup date: expand folders, search a path, tick files or whole folders (a ticked folder covers everything below it), then choose the branch receiving the commit (the source branch or a new branch created from a base branch, with an optional pull request). The rest of the repository is left untouched. The tree is available for backups that captured the code after this feature was released.
Options. Dry run (on by default), destination per repository (the original one, another existing repository in any workspace the token reaches, or a new one), overwrite only with an explicit confirmation, target workspace, repository suffix, project key.
Execution. The worker pushes each chosen ref, writes the files, recreates the configuration item by item (an error on one item never stops the others), logs every step and can be cancelled. Running the same restore again creates no duplicate. Every backup, including an incremental one that reused the previous code bundle, restores the complete state at its date.
Create the target repository when it does not existIn the original project or the one you choose
What cannot be restored identically
Secured variables are recreated empty (Bitbucket never returns their value); webhook secrets must be entered again; only open pull requests are recreated - by the restore account, with the original author and date noted and the comments reposted. Merged and declined pull requests stay available in the export (JSON, CSV, HTML).
You're all set.
Need help? Our team replies within one business day on every plan.