Kodus
An open-source AI reviewer that reads your pull requests for you, checks them against rules you write in plain language, and charges you only what the AI provider actually costs — no markup.
🔗 Visit Kodus
Description
Most AI code review tools bundle the AI cost into a flat subscription, so you're paying a markup on top of whatever OpenAI, Anthropic, or Google actually charge for the underlying model calls. Kodus does it differently: it's open source, lets you plug in the model of your choice, and passes the LLM bill straight through to you at cost, charging separately just for the platform itself.
Kodus is an open-source AI code review tool that integrates with GitHub, GitLab, Bitbucket, and Azure DevOps to automate pull request reviews. Teams write custom review rules in plain language rather than code, and Kodus can validate business logic against connected tools like Jira, Linear, or Notion, not just syntax. It also tracks technical debt and engineering metrics over time, and it's model-agnostic — supporting Claude, GPT-4, Gemini, Llama, and other OpenAI-compatible endpoints — so a team isn't locked into one AI provider's roadmap or pricing.
💬 Our review
The short version: if your team already runs a decent volume of pull requests and doesn't want to pay a markup on every AI review call, Kodus's zero-markup, bring-your-own-model pricing at $10/developer/month is one of the more honest deals in the AI code review space.
Against closed competitors like CodeRabbit or Sourcegraph Cody, Kodus's real advantages are being open source (so you can self-host and audit exactly what it sends to the LLM) and model flexibility — swapping between Claude, GPT-4, or a cheaper open model as pricing or quality shifts, without waiting on a vendor. The tradeoff is that zero-markup pricing means your total cost isn't fixed: it scales with how much you actually use the LLM, which needs a bit more budget discipline than an all-inclusive subscription. Solid choice for teams that already have LLM API relationships and want transparency; less appealing if you'd rather have one predictable monthly bill and never think about token costs.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Gratuit (PR illimitées en auto-hébergé ou hébergé par Kodus, jusqu'à 10 règles, 3 plugins). Teams à 10 $/mois/développeur + coûts LLM directs sans marge. Enterprise sur devis. Essai gratuit de 14 jours incluant jusqu'à 35 revues de PR.
Pros
Open source, auditable, auto-hébergeable
Zéro marge sur les coûts LLM — facturation directe au fournisseur
Agnostique au modèle : Claude, GPT-4, Gemini, Llama, endpoints compatibles OpenAI
Validation de logique métier via Jira/Linear/Notion, pas juste la syntaxe
Cons
Coût total variable selon l'usage réel du LLM, moins prévisible qu'un forfait tout compris
1 300 étoiles GitHub seulement — communauté encore modeste face aux gros concurrents
Nécessite un peu de discipline budgétaire côté coûts de tokens
Date de fondation non communiquée précisément