Capability
Several reviews run over the change, then one decides.
Rather than one general purpose read, separate passes look for logic errors, security weaknesses, performance problems and clarity. A coordinating layer collects what they found, removes duplicates and decides what is worth raising.
What it does
A single general purpose review is pulled in four directions at once and does none of them well. Instead, each pass has one job and reads the change with that job in mind. The logic pass is not distracted by naming. The security pass is not diluted by style.
A coordinating layer then does the work that most tools leave to the reader. It collects everything the passes found, removes the same problem reported four ways, weighs what is left, and decides what is worth your attention.
Effort matches what is at stake
A one line configuration change and a rewrite of the payment path do not deserve the same treatment. Larger and riskier changes get a deeper pass. Small, low risk edits get a proportionate one, which is what keeps both the cost and the noise down.
Why the dedupe matters more than the finding
The most common complaint about AI reviewers is volume. Four passes over a large change will legitimately produce the same observation several times over. Every duplicate that reaches the pull request is a small tax on the reviewer, and enough of them is why teams stop reading.
Four passes, one decision
- Logic 6 raised
- Security 3 raised
- Performance 2 raised
- Clarity 2 raised
13 raised, 11 suppressed
9 were the same problem reported more than once. 2 fell below the confidence bar you set. 2 reached the pull request.
Questions.
How many findings should we expect?
Fewer than you are used to. The coordinating layer suppresses duplicates and anything below the confidence bar, and the assessment reports how many were suppressed so the number is visible rather than hidden.
Can we turn a pass off?
You set the confidence threshold for what reaches you. Per-pass tuning is handled with us during setup so the review stays aligned with the team using it.
See it on your own pull requests.
We install on one repository, let Moonray AI assess the next changes your team opens, and go through what it found with you. If the position is not better than what you have now, you will know within a week.
Nothing to host. Reviews begin on the next pull request opened.