Kodus

About Kodus

Kodus is an open-source AI code reviewer positioned directly against CodeRabbit, and its commercial argument is a pricing structure rather than a feature list: the core is AGPL-licensed and self-hostable for free, and the cloud version runs on your own model API keys, so you pay OpenAI, Anthropic or Google at list price with no markup layered on top. The reviewer, called Kody, comments on pull requests across quality, security and performance, and the piece that distinguishes it from a generic LLM-on-a-diff is Kody Rules — reusable review rules drawn from real codebases that encode a team's conventions, plus a memory layer that carries learnings between reviews so the same nit is not raised forever. A Quality Radar surfaces issues across the repository rather than only in the diff under review. Around the product the team publishes an unusual amount of comparative research: benchmarks scoring AI review tools against real PRs, an LLM performance benchmark on real PR diffs, and a library of agent skills. For teams whose security review asks where the code goes, running the whole thing on their own infrastructure is the default rather than an enterprise upsell. Integration is through GitHub pull requests, and the vendor also publishes a changelog, a customers page and a security page describing how it handles code and keys.

Key Features

Tags