From red test to open pull request.
For a real failure, Cairn drafts a fix with Claude and opens a pull request — closing the loop from detection to diagnosis to a reviewable change.
A diagnosed bug becomes a pull request.
Cairn reads the root cause it found, drafts the fix with Claude, and opens a PR linked to the issue and the run — ready for your review.
Checkout — “card declined” never clears on a valid card
The form treats any non-2xx as a decline and never clears the error.
components/checkout/PaymentForm.tsxFinding the bug is only half the work.
A test that catches a regression has done you a favor — and handed you a chore. Now someone has to read the failure, understand the cause, find the right file, write the fix, and open a PR: all the work the failure itself didn't do for you.
For a large class of bugs that work is mechanical. The cause is clear, the fix is small, and the only thing between a red test and a green one is someone with time to write it.
Auto-fix does that step. When Cairn files a real bug, it drafts the fix with Claude and opens a pull request — so the issue lands with a proposed change you can review, not just a description of the problem.
It writes the fix — you ship it.
Auto-fix works from a real diagnosis and stops when it shouldn't proceed, so what reaches your queue is worth reviewing.
Starts from a diagnosed issue
Auto-fix only runs on confirmed product bugs Cairn has already root-caused — so it works from a real diagnosis and the failing run's evidence, not a vague error.
Drafts the fix with Claude
Cairn hands Claude the root cause, the relevant files, and the trace, and has it write a focused change that addresses the cause — not a band-aid over the symptom.
Opens a pull request
The change arrives as a normal PR against your repo, linked to the issue and the run. Your CI runs, your team reviews, and you merge when you're satisfied.
Knows when to stop
If a fix would need a schema migration or an ambiguous product call, Cairn abandons the attempt and explains why — so you never get a confident, wrong PR.
Closing the loop, without losing control.
Auto-fix proposes; you decide. Every change goes through the same review and protections as code your team writes.
A PR, not just a ticket
Every fixable bug arrives with a proposed change attached, linked to the issue and the failing run — so review starts from a solution, not a blank editor.
You stay in control
Cairn never merges. It opens a pull request and waits for a human; your branch protections, CI, and review process all still apply.
Knows its limits
When a fix needs judgement it can't safely make, Cairn abandons with a reason instead of guessing — keeping misleading PRs out of your queue.
Auto-fix closes the loop that began the moment a flow was mapped: discover → test → run → diagnose → fix. A failing run becomes a reviewed pull request — often before anyone on your team has even opened the issue.
Frequently asked questions
When it files a real product bug. Cairn drafts a fix with Claude and opens a pull request against your repo, so the issue arrives with a proposed change attached — not just a description of what's wrong.
No. Cairn opens a pull request for your team to review and merge. You stay in control — nothing reaches your default branch without a human approving it.
When a fix needs judgement Cairn can't safely make — a schema migration, an ambiguous product decision — it abandons the attempt and tells you why, instead of guessing and opening a misleading PR.
A focused change addressing the root cause Cairn diagnosed, linked back to the issue and the failing run — so a reviewer sees the problem, the proposed fix, and the evidence in one place.
Yes. When a breaking API change is what broke a flow, Cairn can draft a fix targeting that contract change specifically, in addition to UI-level failures.
Like any pull request — read the diff, run it, request changes, or merge. Because the PR links the issue, the run video, and the trace, you review with full context without leaving GitHub.
Explore more of Cairn
Put your tests on autopilot.
Connect a repo and watch Cairn map your flows, write the tests, and run them on every PR — in minutes.