Tool
open-code-review
alibaba/open-code-review
What it is
Reviews code from the command line by reading Git diffs and sending changed files to a configurable language model, through an agent that can open files and search the codebase.
Who it might suit
Teams who want line-level AI review comments on pull requests, using an OpenAI or Anthropic compatible model endpoint they configure.
What to check first
Changed files go to whichever model endpoint is configured, which the README says is separate from any Claude Code subscription. The project reports lower recall than general agents as a deliberate choice, and its benchmark figures are its own.
How it runs
Command-line tool with a model endpoint
Written