OSS
Alibaba's open-code-review shipped v1.12.1 today and claims 4.7x the precision of Claude Code at a fourteenth of the tokens
alibaba/open-code-review is a Go CLI that ran as Alibaba's internal AI review assistant for two years before being open-sourced in May 2026; it pairs deterministic rule pipelines (NPE, thread-safety, XSS, SQL injection) with an LLM agent that can read full files and search the codebase for context. On its own AACR-Bench the project reports 33.90 percent precision with Claude-4.6-Opus versus 7.23 percent for Claude Code on the identical model, at 385K tokens per review instead of 5,664K. It is at 24,587 stars, 1,803 forks and 148 open issues, gained 443 today, and cut v1.12.1 at 11:23 UTC this morning.
↳ Follow the thread
No related signals yet.