Anthropic takes Code Review, the automated code reviewer in Claude Code, out of preview
IA en un minuto newsroom · Editor: Jon Elgezabal
In 30 seconds
Anthropic has ended the trial phase of Code Review, the Claude Code feature that checks pull requests by itself, and is opening it to organizations with a Team or Enterprise plan. It now takes between 6 and 10 minutes in most cases, and each repository can run in Quick, Standard or Deep mode. Enterprise companies that meet the conditions can request credits worth at least $10,000 until January 15, 2027.
Anthropic announced on Thursday, October 15, that Code Review has left research preview and reached general availability. It is the Claude Code feature (Anthropic's AI coding tool) that reviews pull requests, the proposed code changes on GitHub, on its own. Organizations on the Team and Enterprise plans can use it.
The company ties the move to coding agents, which let teams produce code ever faster while review remains a bottleneck. Anthropic says it runs Code Review on its own pull requests, and that teams who tried it in the preview asked for reviews that were less noisy, scaled to the size of each change and quicker.
According to the company, most reviews now finish roughly 3 times sooner than in the preview, dropping from about 20 minutes to between 6 and 10. Findings are shorter, setting out the problem, what causes it and how to fix it in a few sentences. Admins pick one of three levels per repository: Quick, Standard (the default) or Deep.
To set it up, an admin switches it on in the Claude Code settings, installs the GitHub app and chooses the repositories. From then on, reviews start by themselves on every new pull request, with nothing for developers to configure.
Enterprise organizations that qualify have until January 15, 2027 to claim credits for Code Review worth at least $10,000, subject to terms; the credits expire 30 days after being claimed. The announcement does not give a price per review.
Why it matters · analysis and opinion
With general availability, automated review stops being an experiment and becomes part of the regular Claude Code offering, which forces anyone already paying for Team or Enterprise to decide whether they want it in their repositories. Anthropic's argument holds up: if agents multiply the amount of code being written, someone has to read it, and human reviewers do not grow at the same pace. What remains open is the cost, because the announcement does not spell it out and the promotional credits only reach some of the customers. A small team can start with the lightest level on a secondary repository, compare what it finds with what its people already catch and only then bring it to critical projects. The tool does not replace human review: it puts it in order.
Official source: Anthropic · Written with the help of AI: how we make the news
