Open Code Review 🤖 Agent Open source

What it does

Open Code Review (ocr) is an AI code review command-line tool open-sourced by Alibaba, where it began as the company's internal review assistant. It reads a Git diff, groups the changed files into review units and runs an LLM agent over each unit. The result is structured comments tied to exact lines. The design splits the work between fixed code and the model. Ordinary code handles file selection, bundling related files, matching review rules and positioning comments. The agent handles judgement and pulls extra context by reading files and searching the repository. The project says this approach produces fewer false alarms and uses fewer tokens than asking a general-purpose agent to review, though it may find somewhat fewer issues.

What is inside

• Review modes: workspace changes, branch ranges, single commits, resumable sessions and full-file scans with ocr scan
• Rules: custom review rules with path filtering and targeting
• Output: terminal, JSON reports and a browser session viewer for marking comments fixed or ignored
• Delegation mode: OCR selects files and rules, and your own coding agent does the reviewing, so no separate API key is needed
• Integrations: an MCP server for adding external tools, CI recipes for GitHub Actions, GitLab CI, GitFlic CI and Gerrit, and OpenTelemetry tracing

Works with

Plugins are documented for Claude Code (slash commands), Codex, Cursor, Kimi Code and OpenCode. A portable agent skill covers other skills-compatible agents. Any configured LLM provider can run reviews.

How to install

It needs Git 2.41 or newer:

npm install -g @alibaba-group/open-code-review

Then run ocr config provider and ocr config model, followed by ocr review in your project.

Maintenance and safety

The project is Apache-2.0 licensed, maintained by Alibaba and updated often, with an OpenSSF Best Practices badge. Your diffs and any files the agent reads are sent to whichever LLM provider you configure, so choose one your code policies allow. It reads your repository and posts comments. It does not change code.

Who should use it

Teams that want consistent, line-accurate AI review in CI or before opening pull requests, especially on large changesets where general-purpose agents skip files.