Open Code Review - Alibaba's open-source AI code review CLI tool

What is Open Code Review?

Open Code Review is Alibaba's open-source AI code review CLI tool, originating from the group's production-level practices that have served tens of thousands of developers and executed over a million review tasks. The tool employs a hybrid architecture of deterministic engineering and LLM Agent, supporting connections to any LLM and completing reviews locally with completely private data. It includes built-in review rules covering 10+ languages, including Java, Go, and Python. Token costs are only 1/5 of general agent solutions, and it achieved a 26.1% SEM.F1 score in the AACR-Bench benchmark test.

The main functions of Open Code Review

• Hybrid Architecture Review EngineThe deterministic engineering module is responsible for task breakdown, file filtering, line number location, and rule routing; the LLM Agent is responsible for risk detection, context exploration, and problem classification. Each module performs its own function, balancing quality, cost, and speed.
• Precise line-level comment positioningAn independent three-layer progressive LLM positioning module precisely anchors each comment to a specific line number; combined with an independent reflection module, it intercepts illusions and knowledge drift in the early stages.
• Multi-model protocol supportIt natively supports the Anthropic Messages API and the OpenAI Chat Completions API, can connect to custom model endpoints, and is compatible with automatic reading of Claude Code environment variables.
• Dynamic concurrency processingBased on Goroutines, dynamic subtask splitting and parallel review can be completed quickly, even for large change sets.
• Intelligent memory compressionThis solution provides a three-tiered (frozen/compressed/active) context management system for code review scenarios, overcoming token limitations to enable in-depth review.
• Built-in review rulesThe rule base has been verified in a large number of real-world scenarios, covering vulnerability types such as NPE, thread safety, XSS, and SQL injection, and supports custom rule injection with a four-layer priority chain.

How to use Open Code Review

• Installation tools:pass npm i -g @alibaba-group/open-code-review Install globally, or download the corresponding platform binary from GitHub Releases.
• Configuration Model:run ocr config set The command sets the LLM endpoint, token, and model name; if Claude Code is configured, the environment variables will be automatically recognized.
• Verify connectivity:implement ocr llm test The model connection has been confirmed to be normal.
• Start of review:use ocr review Review work area changes,ocr review --from main --to feature Review branch differences, or ocr review --commit abc123 Review individual submissions.
• Integrated WorkflowIt can be integrated into Claude Code as a Skill or Plugin, or via... --format json Output is integrated with GitHub Actions / GitLab CI.

The core advantages of Open Code Review

• Production Validation: Over 20,000 active users within Alibaba, over 1 million review tasks, and an adoption rate exceeding 30%.
• Extremely low costToken consumption is only 1/5 of that of the general Agent + Skills solution.
• Data privacyIt runs locally, and the code does not leave the private environment, making it suitable for sensitive scenarios such as finance and e-commerce.
• Precise positioningThe independent positioning module solves the common "line number drift" problem in general agents.
• Rules matureBuilt-in rules directly address the high-frequency pain points of domestic teams (NPE, SQL injection, XSS).
• Zero-configuration compatibilityAutomatically recognizes Claude Code environment variables, eliminating the need for repeated configuration.

Open Code Review project address

• Project official websitehttps://alibaba.github.io/open-code-review/
• GitHub repositoryhttps://github.com/alibaba/open-code-review

Comparison of Open Code Review with similar competitors

Application scenarios of Open Code Review

• individual developers: Quick self-check before local submission, replacing some manual review steps.
• Platform TeamIntegrate into the internal DevOps system to unify auditing strategies and data flows.
• ML ResearcherAs a code quality validator in the RL training pipeline, it provides reliable reward signals for the code generation model.
• Financial/e-commerce companiesIt operates in an isolated intranet environment, meeting security and compliance requirements while enjoying AI review capabilities.
• Open source project maintainers: Reduce the maintainer's burden by automatically reviewing pull requests through GitHub Actions.