Alibaba/Open-Code-Review

Open Code Review is an AI-powered code review tool developed by Alibaba that uses a hybrid approach combining deterministic engineering with language models to analyze Git diffs and identify code defects with high precision. Originally Alibaba's internal tool serving tens of thousands of developers, it has been open-sourced and outperforms general-purpose AI agents like Claude Code by achieving higher precision and F1 scores while using significantly fewer tokens. The tool addresses common issues with general-purpose agents—such as incomplete file coverage, position drift in reported issues, and unstable quality—through hard constraints on the review process and intelligent file bundling strategies.