Alibaba open-sources OpenCodeReview, an AI code reviewer claiming 9x fewer tokens than Claude Code
thisdudelikesAI · x · 2026-10-27
Alibaba has open-sourced Open Code Review (Apache-2.0, npm v1.13.2), its internal official AI code review assistant. Per the README, it served tens of thousands of Alibaba developers over two years and flagged millions of defects; the project claims 9x fewer tokens than Claude Code with more precise reviews.
Key capabilities:
- Reads Git diffs and produces structured, line-level review comments via a tool-using agent
- Built-in rules for null-pointer bugs, thread-safety, XSS, and SQL injection
- The agent reads full files, searches the codebase, and inspects other changed files for context — not just surface-level diff feedback
- ocr scan reviews entire codebases without a diff, useful for auditing unfamiliar directories
- Plugs into Claude Code, Codex, Cursor, and Kimi Code; supports Windows/macOS/Linux
Free with no paid seats or vendor lock-in — configure a model endpoint and start.
More from coding & agent
- mitsuhiko explains why Pi's classify() only makes sense via codemode with a custom harness — mitsuhiko · 2026-10-28
- AI voice agent for roofing leads: first month set the client's record for booked inspections — NoComment8254 · 2026-10-28
- Theo's T3 Code achieves multi-machine orchestration in a single thread — seatedro · 2026-10-28
- Claude Code voice sessions block /goal until you fork the session — DevDminGod · 2026-10-28
- Five real business use cases for LLM document extraction, from invoices to legal risk summaries — mdancho84 · 2026-10-28
- Google open-sources LangExtract: free LLM document extraction to replace $100K enterprise tools — mdancho84 · 2026-10-28