According to CoinMeta, Anthropic has announced that it is transitioning the automatic code review tool Code Review of Claude Code from a research preview to a formal version, making it available to teams and enterprise users. This feature was launched back in March of this year and allows multiple agent to check the code in parallel after developers submit a GitHub PR (code merge request), in order to identify logical errors, security vulnerabilities, and boundary issues, while also filtering out false positives through cross-validation. Anthropic states that most reviews now take only 6 to 10 minutes, which is nearly three times faster than the approximately 20 minutes of the preview version. Administrators can also set three levels of review depth: fast, standard, and in-depth, for different repositories. The founder of Claude Code, Boris Cherny, stated that he uses Code Review for all his PR projects, and almost all bug issues can be detected before merging. This feature requires additional payment and is not included in the Claude subscription package; the average cost per review is between 15 and 25 US dollars, with actual charges based on the amount of token usage.