Open Code Review runs a deterministic-plus-LLM pass over your git diff to post line-level comments, using Alibaba's internal ruleset and your own model key.
Yes, Open Code Review is open source under the Apache-2.0 license.
Open Code Review is free to use.
Yes, Open Code Review can be self-hosted (the source is available under the Apache-2.0 license).
Apache-2.0 AI code-review CLI that Alibaba ran as its internal official reviewer for roughly two years before releasing it. It reads a git diff, sends the changed files through a hybrid deterministic-pipeline-plus-LLM-agent pass with a built-in multi-language ruleset (null-pointer, thread-safety, XSS, SQL injection), and returns structured, line-level comments. You supply your own OpenAI- or Anthropic-compatible model key, or run a delegation mode that hands the review to a coding agent's own model so no extra key is required. It installs from npm.
A lot of the GTM stack is now code someone on the team vibe-shipped with an agent, and the review step is where that either gets caught or ships. Open Code Review is a self-hostable, bring-your-own-key first-pass reviewer that plugs into the git diff and flags real defect classes at the line level — the kind of thing teams usually rent from a per-seat SaaS. Two honest caveats keep it in proportion: it has been public since May, so it is not the fresh launch some posts imply, and the eye-catching precision numbers are the project's own benchmark, not an independent one. As an Apache-2.0 reviewer with a serious operational history behind it, it is worth wiring in and measuring on your own repo.