Kodus

Engineering teams spend significant time on pull request reviews, with the site's own ROI calculator assuming 30 minutes per review across an organization's PRs. Existing AI code review tools often require enterprise seat minimums, lock teams into proprietary hosting, and mark up the cost of the underlying LLM tokens. Teams also struggle to keep reviews aligned with their own coding standards and business requirements scattered across tools like Jira, Linear and Notion, and to keep track of technical debt from suggestions that never get implemented.

Kodus is an AGPL-licensed, open source AI code review tool that can be self-hosted via Docker Compose or Helm, or used as a cloud service where customers bring their own model keys and pay providers directly at list price. It integrates with GitHub, GitLab, Bitbucket and Azure DevOps at the pull request level, reading diffs and posting inline comments, and is model agnostic, supporting Claude, GPT, Gemini, Llama or any OpenAI-compatible endpoint. Teams can set review rules in plain language or sync existing rule files from Cursor, Copilot or Claude, and Kodus validates PRs against business requirements pulled from Jira, Linear and Notion. It also tracks unimplemented suggestions as technical debt issues over time, and provides an engineering dashboard covering deploy frequency, cycle time, bug ratio and PR size.