Local AI Code Review in CI/CD with Ollama (github.com)

AI Review is an open-source tool by Nikita Filonov that injects AI-powered code review directly into CI/CD pipelines. It supports multiple LLM providers (OpenAI, Gemini, Claude, Ollama, OpenRouter), runs as a CLI or Docker image (pip package xai-review / nikitafilonov/ai-review), and can post inline comments, high-level summaries, cross-file context analyses, and AI-generated replies directly into GitHub, GitLab, Bitbucket, or Gitea merge requests. Review modes include run-inline, run-context, run-summary, run-inline-reply and run-summary-reply, and the repo supplies ready-to-use GitHub Actions and GitLab CI snippets to automate reviews per PR/MR. Technically the tool emphasizes privacy and configurability: it performs all LLM requests client-side (no proxying or storage), reads configuration from .ai-review.yaml/.json/.env, and sends data only to the explicitly configured LLM endpoint—crucially supporting a local Ollama runtime (default http://localhost:11434) for fully offline reviews. Model, temperature, token limits, prompt templates, file inclusion/exclusion and CI overrides are all configurable, and pipelines can be set to not block merges on AI failures. For teams this accelerates review throughput, enforces custom policies via prompt templates, and enables conversational AI participation in existing review threads while keeping source control of data and tokens under the organization’s control.