Self-host for free, or use our cloud with your own model keys.
You pay the model provider at list price, zero markup.
14-day free trial • up to 35 PR reviews included • no credit card required
Trusted by 5,000+ teams in 66 countries
The Kodus core is open source under the AGPL license.
You can read the review logic, audit what touches your code, and run it on your own infrastructure with Docker Compose or Helm.
No black box between your code and production.
Kodus doesn’t cap how many reviews you get per hour.
With your own API key, every plan reviews unlimited PRs. Push, rebase or force-push as often as you need, and each change gets reviewed.
Bring your own API keys on every plan, cloud included. Use any provider with an OpenAI-compatible API, pick a different model per repository, and set a fallback model. You pay for tokens directly to your provider, at list price.
On the Teams plan, the $10 per user is strictly for platform infrastructure. Your model spend stays on your own bill, with the provider you choose.
Run Kodus on your own infrastructure without an enterprise contract.
Clone the repository and deploy with Docker Compose on a VM, or with Helm on Kubernetes. Point it at your Git provider and your model keys, and reviews stay inside your network.
Self-hosting is available on the free Community plan. Enterprise adds SSO, RBAC, audit logs and dedicated support when you need them.
You want a poem in every pull request.
You want one vendor picking your models and marking up every token.
You think every team should review by its own rules.
You enjoy 50 auto-generated comments on every pull request.
You like waiting for your review quota to reset before you can push again.
Kody maps the callers, callees and tests around every change, including code in the other repos you link.
Write review rules in plain language, or import the ones you already keep in .cursorrules, CLAUDE.md or AGENTS.md.
Kody pulls requirements from Jira, Linear and Notion, then checks every PR against them, and flags what breaks.
Ask why something was flagged, or push back. When you correct Kody, it offers to save that as a memory for the next reviews.
Unimplemented suggestions become issues automatically, so debt stays visible and shrinks over time.
Deploy frequency, cycle time, bug ratio and PR size in one dashboard, all trending the right way.
Luiz Barrile
@Lerian
Kodus has become an essential part of our process at Lerian. By standardizing steps and automating checks, we’ve gained more speed and consistency, while reducing rework and improving delivery quality.
Pedro Maia
@Notificações Inteligentes
We trained the team to use AI in day-to-day coding, and Kodus stepped in as our senior reviewer that never forgets anything. It doesn’t replace human review, but it’s now a required step: it ensures consistency and prevents repeat incidents.
Oleksandr Kuchma
@SaaSJet
We love what Kodus does. It has dramatically reduced our PR review time, and our developers no longer want to review a PR without Kodus running first. The accuracy is very good: it catches many of the small issues that are easy to miss, allowing our developers to focus on the architectural decisions that truly require human judgment.
Ricardo
@Ikatec
Since we started using Kody, the dev experience has improved a lot. Time spent on code reviews dropped by around 30%, and the AI brings valuable insights on performance, security, and code optimization. One of the best parts is that we can tailor how it works for each project.
Raphael Sampaio
@Pilar
Kodus has been helping us save a lot of time on code reviews, while also providing key engineering productivity metrics. Since we started using the tool, our average review time has dropped from hours to minutes.
Jonathan Georgeu
@Origen
Kodus has had a huge impact on our workflow by saving us valuable time during PR reviews. It consistently catches the small details that are easy to miss, and the ability to set up custom rules means we can align automated reviews with our own standards.
João H. Kersul
@Doji
These days, Kodus is part of our daily review routine. It helps a lot with error handling and brings up suggestions that would often go unnoticed. This active listening and fast turnaround have made a real difference for our engineering team.
Igor Duca
@ducaswtf
Kodus helped me move as fast as I ever could during my development days. It has never been so easy to ship reliable code and build real solutions.
André Diogo
@Brendi
Kodus fit like a glove for me. Before, I was buried in slow code reviews. Now, feedback happens way faster, and I can actually focus on other things.
David Barnett
@QuintoAndar
Kodus helps us reflect our standards in PRs to share knowledge and raise our code quality. Kody catches some subtle issues and calls attention to them so reviews and authors can have a more effective review. I appreciate the flexibility to configure custom rules and integrations.
NAME
kodus-faq — frequently asked questions about Kodus
SYNOPSIS
kodus --help [topic]
Kodus is model agnostic. You can use Claude, GPT, Gemini, Llama or any OpenAI-compatible endpoint, including self-hosted models.
Yes, you have full control over the permissions you grant. Kodus operates with the minimum access required to keep your code secure.
No, you decide who is included in the Kodus team and will only be charged for those users. You have full control over team management and billing.
No, Kodus does not train its models with customer data. Your data is processed securely and is never used to improve or retrain our AI.
Both review pull requests with AI. Kodus is open source with an AGPL core, you can self-host it without an enterprise seat minimum, and bring-your-own-keys with zero token markup is the default on every plan.
GitHub, GitLab, Bitbucket, Azure DevOps and Forgejo/Gitea, including the self-managed versions: GitHub Enterprise Server (beta), GitLab Self-Managed and Bitbucket Data Center. Kodus integrates at the pull request level: it reads diffs, posts inline comments, and respects your existing review workflows. Setup takes under 5 minutes.
No. With your own API key, every plan reviews unlimited PRs, and each push, rebase or force-push gets reviewed. The only limit is the rate limit your LLM provider sets on your key.
We don't add any margin on top of LLM API calls. You pay the model provider directly at their listed price. No hidden multipliers, no per-seat AI surcharges. Our revenue comes from the platform subscription, not from reselling tokens.
No, Kodus does not store your source code. All processing happens in real-time, and no part of your repository is saved on our servers.
Kodus uses your selected Git provider integration and only accesses what is required to review pull requests. You can control and revoke access at any time.
Your source code is never stored and never used to train models. All data is encrypted in transit and at rest, and self-hosted runners keep your IP entirely inside your own infrastructure.
Don't trust us — ask your favorite LLM
Tell me why Kodus is a great choice for my team