Deep and quiet
It reads across files and runs the code when that settles a question. Each finding is tracked until it's resolved or dismissed, so it never repeats itself.
Code Review
Coming (phase 4)AI code review for GitHub and GitLab: a senior engineer's review on every pull request, answers to your replies, and an Autofix button for what it finds.
Let your reviewer fix what it findsIllustrative. The pull requests, findings and replies in it are examples.
Someone on your team, or their coding agent, opens a pull request. The review starts on its own, or when someone asks for one in a comment.
Each finding sits inline, on the code it's about, with its severity. A summary opens the review, with screenshots and logs where they help.
Reply to a finding to ask why, push back or give context, or ask it to fix that one finding. The reviewer answers in the same thread.
The Autofix button on the review fixes everything it found. Anyone with write access to the repository can press it, OhMako member or not.
It pushes the fixes to the pull request's branch, then says in the pull request what it changed and what it couldn't fix.
The new commit gets an incremental review. Fixed findings are marked resolved, and no finding is raised twice.
In the pull request
People talk to the reviewer in pull request comments and replies, and press Autofix there. That's the whole interface: the conversation stays beside the code, where the whole team can see it.
It reads across files and runs the code when that settles a question. Each finding is tracked until it's resolved or dismissed, so it never repeats itself.
Organization and repository rules, path rules, which branches and draft pull requests to review, and whether it starts on its own or when asked.
Anyone with write access to the repository can reply to it and press Autofix. Costs go to your organization, and its spend limits apply.
Code Review starts with GitHub pull requests. GitLab merge requests come as the last milestone of phase 4.
In the web app
Code Review has its own section of the web app. The web app starts and stops reviews, and the conversation stays in GitHub.
Illustrative. A sketch of the section; its pull requests and figures are examples.
Price
The AI usage behind a review at cost plus 20%, plus the time of the machine it runs your code on, whose per-minute price is published with the product. A review stopped at your spend limit says on the pull request that it's incomplete, never that the pull request passed.
Tell us which products you want first. Invites go out in waves, product by product.
Let your reviewer fix what it finds