Launched this week
Revix AI
AI code review that learns your repo's own rules
115 followers
AI code review that learns your repo's own rules
115 followers
Revix AI scans your repo, generates its own pattern and architecture rules, and reviews every pull request against them. It keeps learning as the repo changes. Each comment names its rule and is labeled by intent and severity. Free for public repos.
Free Options
Launch Team / Built With
CloudflareDeploy serverless functions globally in seconds
Promoted
Hey Product Hunt 👋
I'm Uri. Oscar and I are building Revix AI from Barcelona.
Why we built it: LLMs made us write code much faster, but we watched quality drop and incidents go up. Review is where that should get caught, and it usually doesn't, because most review comments don't know how your repo actually does things. So people stop reading them.
Revix starts by scanning your repo and generating its own pattern and architecture rules. You don't write a rulebook. It reviews every pull request against those rules and keeps learning as the repo changes.
Each comment names the rule behind it and is labeled by intent and severity, so you know why it's there and what to fix first. The number we care about is how many comments you act on, not how many we leave.
It works with GitHub, GitLab and Bitbucket Cloud. It's free for public repos, with a 14 day trial and no card for private ones.
It's early and some comments will miss. Please be blunt: run it on a repo you know well and tell us which comment was useless and why. Oscar and I are here all day.
review that learns the repo's own rules is usefull, most tools are so generic 👏 how fast does it pick them up?
@petrkovacik it takes just a few minutes to get all of them, depends on how big is the repository. Normal repos are around 1-3 minutes. Then the rule updates after merging PRs is almost instant. Reviews take around a minute, depending as well on how big it is.
where does it learn the rules from? if it’s past review comments, wouldn’t it pick up the ones that were wrong too?
@inhwa The first step Revix AI does is analyze the whole repository, extracting all the standards and patterns into structured rules. Then those rules are used (paired to other mechanisms) to perform the review.
While you update your repository, Revix is clever enough to add, delete or modify existing rules in order to adapt to the reality of the repository after every pull request. It knows what is an ignored issue (and should not modify any rule) or what is a change in the standards of the repository that deserves a new rule.
Naming the rule behind each comment is the detail that matters most to me — a review comment without the 'why' is easy to ignore. How do learned rules get retired when the codebase moves on? I'd worry about an old pattern sticking around as a rule long after the team stopped following it, and having to argue with the reviewer about it in every PR.
@jcbuilds Whenever a pull request is merged, Revix AI analyzes the changes introduced and any human comments. Then it decides when it is needed to add, delete, or modify a rule, so the reviews are always synced to the reality of the repository. We've put a lot of effort into making Revix completely plug and play, with no configuration needed and self-updating, because our goal is to make you win more time for coding and spend less on reviewing or updating documentation.
Hello there,
I'm Oscar, the other co-founder.
I wanted to explain quickly the story behind Revix AI:
Uri and I found that in this AI era we were reviewing more than coding in our team, and that mostly ended up with us not being able to carefully check everything in a PR, just a quick look at each one. Of course we then tried other AI code reviewers out there, but we could not trust any of them. They just added so much noise that other devs stopped reading the bot comments. We needed a 1 to 1 replacement for us, an AI senior reviewer that matches our style and let us focus on the more human things about coding.
Once we achieved that internally, we thought this was too good not to share with everyone, and that is why we put tons of hours into the afternoons and weekends to make it even better, till it became our main project. Today we are announcing it to the world, and we cannot be more proud of what we've made.
Hope you guys like it as much as we do!
Last year Claude Code went through a few weeks of ignoring CLAUDE.md, so a rule being written down doesn't mean it gets followed. Checking every PR against the repo's own rules, and naming the rule in each comment, is a much better place to catch it.
@simon_moxon Revix AI has some mechanisms to ensure this happens. If your project have clear standards in the CLAUDE.md file, those will become rules that Revix then will use for the review whenever they must apply.