Bitbucket integration

Bitbucket Integration

Connect a Bitbucket workspace once and every pull request in the repositories you pick gets reviewed. Add the IronBee Pipe and each one is verified against your running app, with the verdict on the pull request itself.

Connect Bitbucket. Every pull request gets reviewed and verified.

  1. A pull request opensOr gets a push. IronBee picks it up on its own, with no one asking.
  2. Reviewed on its linesCode review reads the change with your team's rules and memories, and comments on each finding with a suggestion.
  3. Verified in a real browserThe IronBee Pipe drives the change against your real backend, on IronBee or in your own pipeline.
  4. Both on the pull requestThe review and the verification each post their findings and a build status. The evidence is in the Console.

Connect the workspace once. From then on every pull request in the repositories you pick is reviewed, the IronBee Pipe verifies it in Bitbucket Pipelines, and the evidence lives in the Console.

Code review

A review on the lines of the change

IronBee comments on each finding with a suggestion you can apply, adds a review comment and a summary comment, and sets the IronBee review build status.

Let customers upload documents #96document-upload into mainDiff
documents/views.py+14−3
18def upload(request):
19 f = request.FILES["file"]
20 name = f.name
20 name = f.name.encode("ascii").decode()
ironbee-aiBotreviewed 3 hours ago
HighCorrectness

Encoding the file name as ASCII fails on any accented letter, so those uploads end in an error instead of being saved.

Happens when: someone uploads a file whose name has an accented letter, such as Résumé.pdf.

Suggested change
name = f.name.encode("ascii").decode()
name = get_valid_filename(f.name)
Apply suggestion

Rule unicode-input what people type is stored as they typed it, in any language.

21 return Document.objects.create(owner=request.user, name=name, file=f)
IronBee review2 findings · 1 high · 1 mediumFailed
Verification

The whole run, on the pull request

With the IronBee Pipe, the run lands in the same pull request: what the agent did, what it found, and where it came from, as a summary comment and inline comments.

IronBee VerificationVerification failed · 2 findings2m 58sFailed
ironbee-aiBotcommented 3 hours ago
Open in IronBee

IronBee Verification

Verification failed1 high · 1 medium

SeverityFindingWhere
HighUploading Résumé.pdf ends on a 500 error pagedocuments/views.py:20
MediumA 30 MB file is accepted, though the page says the limit is 10 MBdocuments/forms.py
Checks2 passed · 2 failed
  • Signed in and uploaded notes.pdf
  • Uploaded Résumé.pdf
  • Uploaded a 30 MB file past the 10 MB limit
  • Opened notes.pdf from the documents list
Summary
  • A file name with an accented letter fails to upload, and the 10 MB limit the page states is not enforced.
Important files changed
  • documents/views.py
  • documents/forms.py
  • templates/documents/list.html

View the full verification →Used 3 IronBee credits.

Access

An app on your workspace, never your name

IronBee is an app on Atlassian's Forge platform. A workspace admin installs it once, and it works on the repositories you pick.

01

Its own app user

Everything IronBee writes on Bitbucket appears as its app. It uses no seat, needs no webhooks, and you create no token for it.

02

No write access to your code

The app reads repositories and pull requests, and writes on pull requests only. Fixes are pushed by the IronBee Pipe, with the step's own token.

03

Merges held by your merge check

IronBee's build statuses count toward Bitbucket's merge check on successful builds. On a Premium plan, where an admin turns it on, that check holds the merge.

Bitbucket FAQ

Reviewing and verifying Bitbucket pull requests, answered.

How the Bitbucket connection and the IronBee Pipe review and verify every pull request.

A workspace admin installs the IronBee app on the workspace, connects it to your IronBee account from the app’s page in the workspace settings, and picks the repositories. IronBee then reviews every pull request in them: an inline comment on each finding with a suggested fix, a review comment, a summary comment and the IronBee review build status. Mention @ironbee-ai on a pull request to ask for a review.

Now generally available

Ready to ship
AI-generated code with confidence?

Create your account and start verifying what your agents ship. Catch regressions before they reach production.

No credit card required.