← Back to Discover

Kodus

Open-source AI code review for pull requests that runs on your own model keys. Self-host it or use the cloud, with plain-language review rules, issue tracking and delivery metrics.

Code review
Visit website ↗

What it does

Kodus reviews pull requests with AI. Its core is released under the AGPL, so the review logic can be read, and it can be self-hosted with Docker Compose or Helm or used as a hosted cloud service. On every plan you supply your own model key and settle token costs with the provider at its listed price. Its reviewer, Kody, looks at what calls and is called by each changed function and at related tests, including code in other repositories you link. Teams express review rules as ordinary sentences or bring in rules they already keep in .cursorrules, CLAUDE.md or AGENTS.md. Kody can also compare a pull request with task requirements held in Jira, Linear or Notion. In the pull request you can ask Kody why a comment was made, and it offers to remember your corrections for later reviews. Suggestions left unaddressed when a pull request closes become tracked issues, and a dashboard reports delivery measures such as deploy frequency and PR cycle time. Supported Git hosts include GitHub, GitLab, Bitbucket, Azure DevOps and Forgejo/Gitea, with some self-managed editions. According to the site, Kodus does not store source code or train models on customer data.

Capabilities

  • Open-source AGPL core that can be self-hosted with Docker Compose or Helm
  • Bring your own model key, including any OpenAI-compatible provider, with no token markup
  • Review rules written as plain sentences, with imports from .cursorrules, CLAUDE.md and AGENTS.md
  • Cross-file and cross-repository context for each change
  • Compares pull requests with task requirements stored in Jira, Linear or Notion
  • Unimplemented suggestions tracked as issues, plus a delivery metrics dashboard

When to use it

  • Your team wants AI review on pull requests while choosing and paying for the model itself
  • You need code review to run on your own infrastructure
  • You want team-specific review rules enforced automatically on every push
  • You want to track unaddressed review suggestions and delivery metrics

Alternatives to explore

  • CodeRabbit