Kodus
Open-source AI code review platform with BYOK and zero token markup.
About
Kodus is an open-source AI code review platform (kodustech/kodus-ai, AGPL-3.0 core with dual commercial licensing for enterprise features, around 1.3k stars) that works on GitHub, GitLab, Bitbucket, and Azure DevOps. It is model-agnostic: bring your own key for Claude, GPT, Gemini, Llama, GLM, Kimi, or any OpenAI-compatible endpoint including self-hosted models for a fully air-gapped pipeline, and you pay the model provider directly with zero markup plus a token-usage dashboard.
Review rules (Kody Rules) are written in plain language and inherit from global to repository to directory scope, and Kodus auto-detects existing rule files like .cursorrules, CLAUDE.md, AGENTS.md, and copilot-instructions so the standards your coding agents follow are enforced at review time. Linked repositories let the reviewer read sibling repos to catch cross-repo contract breaks. Limitations: the AGPL-3.0 license requires care for companies modifying and distributing the code, it is a smaller project than commercial competitors so ecosystem and support are thinner, and self-hosting needs Docker Compose or Kubernetes operational effort.
What it can do
- AI PR reviews with repository and cross-repo context
- Plain-language Kody Rules synced from existing config files
- Bring-your-own-key with zero token markup
- Self-hosted Docker Compose or Kubernetes deployment
- Supports local models via OpenAI-compatible endpoints
Setup
Deploy via Docker Compose on your own VM or Kubernetes via Helm, connect your Git provider, and add your model provider API key.
Prerequisites
A server or Kubernetes cluster for self-hosting; a Git provider integration; an LLM API key.
Use cases
Works with
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 68; Trending score 62/100 from researcher assessment.
★ 1,300 GitHub stars