OpenCodeReview

OpenCodeReview is an AI code review CLI that started as Alibaba Group's internal review assistant. It reads Git diffs, sends the changed files to a configurable LLM through an agent with tool use, and returns structured review comments anchored to exact lines. The agent can read full files, search the codebase, and inspect other changed files for context. A full-file scan mode reviews whole files or directories that have no meaningful diff, for auditing unfamiliar code.

The design splits the review between deterministic code and the agent. Deterministic code selects which files need review, bundles related files into review units that each run as a subagent with isolated context, matches review rules to each file through a template engine, and runs separate modules that check comment position and content. The agent handles dynamic decisions and context retrieval, using prompts and a toolset tuned on tool-call traces from production review data.

On AACR-Bench, a benchmark of 200 real pull requests from 50 open-source repositories in 10 languages with 1,505 annotated issues, the project reports higher precision and F1 than Claude Code with the same underlying model, at about one ninth of the tokens and with shorter review times. Recall is lower, a trade-off chosen in favor of fewer false alarms.

Reviews run with an LLM that OpenCodeReview manages, or in delegation mode, where OpenCodeReview handles file selection and rule resolution and your coding agent performs the review with its own model.