Claude: Code Review is now generally available

Anthropic has taken Code Review out of research preview and made it generally available to Claude Team and Enterprise organizations, claude.com's report says. Eligible Claude Enterprise organizations can claim at least $10,000 in Code Review credits by January 15, 2027. Those credits expire 30 days after the organization claims them, and terms apply. The credit names eligible Enterprise organizations. Team plans get the generally available review.

This piece covers the speed, quality, and flexibility changes Anthropic shipped, how admins turn reviews on, and the credit deadline with its 30-day expiry. It is for Claude Team and Enterprise admins who handle Claude Code settings and the GitHub App, and for developers whose new pull requests are reviewed after a repository is selected. Once an admin enables Code Review, developers add no configuration of their own.

What Code Review is now generally available shipped

Anthropic points to coding agents that keep raising the pace at which teams write code, and to a review bottleneck that has not eased. For general availability, the company made Code Review faster, better at finding the bugs that matter, and easier to tune to each repository. Since the research preview, Anthropic says Code Review has gotten sharper and catches more of the bugs that would otherwise reach production. Claude Team and Enterprise organizations are the plans that now have this release.

Anthropic runs Code Review on pull requests inside the company and says that use shows where reviews slow its own engineers down. Teams that used the research preview added the rest. They told Anthropic it was catching real issues before a human reviewer looked at the pull request, and that they wanted reviews that were quieter, sized to each change, and faster. The changes in general availability follow that feedback. Anthropic names the three pieces it shipped: speed, quality, and flexibility.

What changed for builders in Code Review is now generally available

On speed, Anthropic says most reviews now finish about 3x faster than during the research preview, from about 20 minutes down to 6-10 minutes. On quality, Code Review finds more of the bugs that matter and returns clear, concise findings. A finding states the problem, what triggers it, and the fix, in a few sentences. Anthropic places that sharper catch rate in the period since the research preview, when Code Review caught more of the bugs that would otherwise reach production.

On flexibility, admins choose Quick, Standard, or Deep for each repository, and Standard is the default. Routine codebases get a quick pass. The most critical repositories get a deep review. Anthropic says every level delivers high-quality reviews. The per-repository choice is how an admin sizes the review, which is what preview teams asked for when they wanted reviews sized to each change. They also asked for quieter reviews. Findings stay short: the problem, what triggers it, and the fix, in a few sentences.

How to install or upgrade Code Review is now generally available

Code Review is generally available for Team and Enterprise plans. Admins enable Code Review in Claude Code settings, install the GitHub App, and select the repositories that should receive reviews. The GitHub App is the integration those steps name, and the repository list is where an admin limits the run. On each selected repository the admin sets Quick, Standard, or Deep, with Standard as the default. A routine codebase can take the quick pass, and the most critical repository can take the deep review.

Developers have no setup of their own. Once an admin enables Code Review, Anthropic says reviews run automatically on new pull requests and no configuration is needed. Those reviews apply to new pull requests on the repositories an admin selected. Team and Enterprise organizations share that path. Eligible Claude Enterprise organizations can claim at least $10,000 in Code Review credits by January 15, 2027. Credits expire 30 days after the claim, and terms apply. GitHub App setup stays with admins.

Gotchas and compatibility in Code Review is now generally available

The credit reaches fewer organizations than the product does. Code Review is generally available to Claude Team and Enterprise organizations. At least $10,000 in Code Review credits is for eligible Claude Enterprise organizations, and the claim deadline is January 15, 2027. Credits expire 30 days after the organization claims them, so a claim on that deadline still lasts 30 days from the claim. Anthropic says terms apply. The stated credit is at least $10,000. The credit line does not name Team.

A developer cannot start Code Review alone. An admin enables it in Claude Code settings, installs the GitHub App, and selects repositories before automatic reviews begin. Anthropic describes those automatic reviews for new pull requests and says developers need no configuration. Depth is set per repository: Quick, Standard, or Deep, with Standard as the default. Routine codebases get the quick pass, and the most critical codebases get the deep review. Anthropic says every level delivers high-quality reviews for that repository.

What to watch after Code Review is now generally available

Eligible Claude Enterprise organizations can claim at least $10,000 in Code Review credits by January 15, 2027, and the credits expire 30 days after the claim. Terms apply. The deadline does not extend those 30 days. Claude Team organizations have the general-availability product and are not named for the credit. Anthropic also runs Code Review on its own pull requests and says that use shows where reviews slow its own engineers down, next to the preview request for quieter and faster reviews.

Most reviews now finish about 3x faster than the research preview, from about 20 minutes down to 6-10 minutes. Admins still choose Quick, Standard, or Deep on each repository. Standard is the default, and every level delivers high-quality reviews. A finding uses a few sentences to state the problem, what triggers it, and the fix. Preview teams said real issues were caught before a human reviewer looked at the pull request, and they asked for reviews that were quieter, sized to each change, and faster.