Open Code Review

Open Code Review (ocr) is Alibaba Group’s internal AI review assistant, open-sourced after years on tens of thousands of developers. It reads git diffs (or full trees via ocr scan), runs a hybrid stack of fixed engineering steps and a tool-using agent, and emits structured line-level comments. Built-in rules cover NPE, thread-safety, XSS, SQL injection and more across many languages; providers are OpenAI- and Anthropic-compatible.

Coding agents ship more diffs than humans can review. Pure agent skills miss files, drift line numbers, and burn tokens. Open Code Review hard-codes selection, bundling, and positioning so the model only does judgment — higher precision at roughly one-ninth the tokens versus a general agent on their bench.

Install with npm i -g @alibaba-group/open-code-review, then ocr config provider and ocr config model. In a repo run ocr review for workspace changes, ocr review --from main --to feature for a branch range, or ocr scan for whole-file audits. Use --format json for CI and host agents; ocr delegate lets Claude Code, Codex, or Cursor run the review with OCR still owning file pick and rules. Plugins and MCP extend the agent; GitHub Actions and other CI docs ship on the site.

Not another free-form review prompt and not a diff-reading UI for humans. The product is the hybrid harness: engineering guarantees coverage and comment placement; the agent does deep context. That is a different job from code-review graphs or “make the PR readable” tools.

• Public demo available
• MCP server / client support

• meat
• Code Review Graph
• Deepsec