[GitHub Trending] alibaba/open-code-review
Scored daily by a customisable AI persona to surface the most relevant engineering leadership news.
Alibaba's hybrid code review tool with LLM agent is highly technical and actionable.
Alibaba open-sourced OpenCodeReview, an AI-powered CLI tool that reads Git diffs and sends changed files to a configurable LLM agent for line-level code review. Its hybrid architecture combines deterministic engineering (precise file selection, smart bundling, rule matching) with an agent, achieving higher precision and F1 than general-purpose agents like Claude Code while consuming ~1/9 the tokens. The tool includes an `ocr scan` mode for full-file auditing and is validated against AACR-Bench, a real-world benchmark of 200 PRs across 10 languages with 1,505 annotated issues.