Show HN: AI Review – Universal AI Code Review for Any LLM or CI/CD (github.com)

Over the last few weeks, I’ve been building AI Review — an open-source framework that turns any CI/CD pipeline into a fully automated AI-powered code reviewer.

Unlike typical SaaS bots, AI Review doesn’t send your code anywhere. It runs entirely within your infrastructure, using whichever LLM or VCS you prefer.

Supports any LLM provider — OpenAI, Claude, Gemini, Ollama, OpenRouter, or your own inference server. Works with any VCS — GitHub, GitLab, Bitbucket, Gitea. Runs anywhere — Docker, GitHub Actions, GitLab CI, Jenkins, or locally.

The setup takes 15–30 minutes. It automatically analyzes diffs, posts inline comments, summaries, and even AI-powered replies inside your PRs/MRs.

Example:

ai-review: image: nikitafilonov/ai-review:latest script: - ai-review run variables: LLM__PROVIDER: "OPENROUTER" LLM__META__MODEL: "anthropic/claude-3.5-sonnet" VCS__PROVIDER: "GITLAB" VCS__PIPELINE__MERGE_REQUEST_ID: "$CI_MERGE_REQUEST_IID"

It’s built around modular prompts and review contracts — you can tune depth, style, or rules for each project.

In short, AI Review is a plug-and-play AI reviewer for real-world teams — no tokens, no vendor lock-in, no data leaves your infra.

- GitHub: https://github.com/Nikita-Filonov/ai-review - Docs & examples: https://github.com/Nikita-Filonov/ai-review/tree/main/docs - Feedback: https://t.me/nikitafilonov

Feedback welcome — I’d love to hear how others handle AI code review in enterprise CI/CD setups.