FileConcat

Bitbucket import

Paste a public Bitbucket repository URL into the import panel and press Fetch, and FileConcat fetches the project tree.

Supported URL shapes

The Bitbucket adapter at packages/core/src/sources/adapters/bitbucket.ts matches bitbucket.org URLs only. Bitbucket Server and Data Center (the self-hosted products) are not supported.

https://bitbucket.org/workspace/repo
https://bitbucket.org/workspace/repo/src/branch
https://bitbucket.org/workspace/repo/src/branch/subdirectory

A trailing .git is tolerated and stripped. When a subdirectory is included, only files at or under that path are imported.

The Bitbucket workspace name fills the same slot as owner does for GitHub and GitLab URLs.

The ref is resolved to a commit first

Before listing anything, the adapter requests the repository's /src endpoint and reads the commit hash out of the URL it was redirected to (resolveBitbucketRef). Only then does it walk the tree, at that commit.

That extra request buys two things the other two sources do not have:

  • The default branch is whatever the repository actually uses. Nothing is guessed and nothing falls back to main.
  • Branch names containing a slash work. src/release/2.0 resolves through the redirect instead of being split into a branch and a subdirectory, which is what happens on GitHub import and GitLab import.

What one import costs in API requests

Bitbucket has no single recursive-tree endpoint, so the adapter walks directories. For one import:

  • One request to resolve the ref.
  • One request per directory, plus one more for each 100 entries inside it.
  • One request per file, twelve at a time.

Every one of those goes through api.bitbucket.org. Atlassian documents a strict limit of 60 requests per hour for anonymous callers, across all API resources (Atlassian, API request limits, read September 2026).

Put the two together and the arithmetic is the thing to know before you paste a URL: a 40-file project spread over 8 directories costs about 49 requests, which is most of one hour's anonymous budget. Bitbucket is the source here where repository size decides whether the import finishes, and the ceiling is a few dozen files rather than a few thousand. Point the URL at a subdirectory to stay under it, or download the project and use the drag-and-drop path instead.

When a request is denied or a file fails

Errors surface through the same classifyResponseError helper used by the GitHub and GitLab adapters (packages/core/src/sources/adapters/_errors.ts), with distinct wording for 401, 403 and 429 responses. A 429 whose Retry-After hint fits inside 30 seconds is retried once automatically.

A file that fails to download does not fail the import. The failure is recorded with its path and reason and the rest of the bundle is assembled without it, so check the file count in the result against what you expected.

Public repositories work without authentication. Private repositories are not accessible at the time of writing; token-authenticated imports are tracked as separate work.

See also