中/EN
Flash

TopicTrending open-source AI on GitHub

Flash·Apps & Agents·2026-10-10 08:06

Alibaba Open-Sources Its Internal AI Code Review Tool

Alibaba Group has open-sourced its internal AI code review assistant, OpenCodeReview. Over the past two years it served tens of thousands of developers inside Alibaba and caught millions of code defects; now released under Apache 2.0 on GitHub, it hit the trending charts on October 9 with stars passing 45,000. It uses a "deterministic engineering × LLM agent" hybrid architecture: engineering logic handles the fixed process steps, while the agent handles only dynamic decisions and context retrieval.

HIMIAO Horizon·2026-10-10 08:06·1 min read

Listen Save Share card Copy link
An engineer pointing at a screen showing AI code review results with green checkmarks
OpenCodeReview passes 45K stars

What happened

On October 9, alibaba/open-code-review hit GitHub trending, with stars passing 45,000. It is the open-source version of Alibaba's official internal AI code review assistant. Per the README, it served tens of thousands of developers and caught millions of defects over two years before being incubated as open source. One command to install — npm install -g @alibaba-group/open-code-review — then ocr review reviews diffs and ocr scan audits whole repos, with a configurable model endpoint compatible with OpenAI and Anthropic APIs.

Key facts

  1. Hybrid architecture:deterministic engineering handles the must-not-fail steps (file selection, smart bundling, rule matching, comment positioning and reflection); the agent handles only dynamic decisions and context retrieval.
  2. Benchmarks:on its own AACR-Bench (50 popular repos, 200 real PRs, 10 languages, 1,505 issues annotated by 80+ senior engineers), it beats general-purpose agents on precision and F1 with the same underlying model while using ~1/9 of the tokens — at the cost of slightly lower recall.
  3. Batteries included:GitHub Actions / GitLab CI / Gerrit integration, a session-replay browser, OpenTelemetry observability, and a delegation mode that lets your own coding agent perform the review.
  4. Apache 2.0:45,182 stars at press time, 137 releases, with ecosystem plugins for Claude Code, Codex, Cursor, Kimi Code and other major coding agents.

Why it matters

It ran inside Alibaba for two years and caught millions of defects; on its own AACR-Bench it beats general-purpose agents on precision and F1 with the same model while using about 1/9 of the tokens.
Useful Tap if this story helped you

CompaniesAlibaba

SourcesGitHub 仓库页 (2026-10-09, source). This article is compiled from public information and does not constitute investment advice.

From HIMIAO Horizon · himiao.ai

ALIBABACODE-REVIEWOPEN-SOURCE

Comments

  1. Loading comments…

Related

  1. Oct 5Open-Source Strata Runs a 125-Billion-Parameter Model on a 12GB Gaming PC
  2. Oct 19Jailbroken PS5 Becomes a Local AI Server: Qwen3.8-27B Hits ~20 tok/s, Small Models Top 100
  3. Oct 15Apple Intelligence Launches in China in December with iOS 27.3, Excluding New Siri AI: Report
  4. Oct 13Alibaba's Qwen Introduces QwenGyre: Elastic RL for Extreme-Long-Horizon Agents
  5. Oct 10Qwen-Image-2.1-Turbo: High-Speed Image Generation in 8 Steps

Today

Oct 19 Monday

Latest news

All →
  1. Oct 18Anthropic to Commit 5% of Shares to New Public Benefit Fund
  2. Oct 19Stably AI's Orca Tops GitHub Daily Trending: One Desktop App to Orchestrate 30+ Coding Agents
  3. Oct 19Anthropic Internal Study: Junior Share of Technical Staff Falls to 14%, Meeting Time Up 20%
  4. Oct 18Claude Cracks 25-Year Cryptography Mystery in Half an Hour: NIST Curve Seeds Came From a 1997 Clock
  5. Oct 18OpenCode Hit by "Distillation Attack": API Keys of About 1.4% of Users Compromised
  6. Oct 18Websites Are Closing Their Doors to AI Agents: Yelp, Delta, JPMorgan Draw Lines
Ask the cat