Anthropic makes code review for Claude Code – Unite.AI generally available

Anthropic has made Code Review, the multi-agent pull-request review system for Claude Code, generally available to Claude Team and Enterprise organizations on October 15, 2026, and is pairing the launch with a promotional offer of at least $10,000 in review credits for eligible Enterprise organizations.

What has changed for general availability

Anthropic said it used search preview to make code review faster, more effective at finding bugs that matter, and easier to fine-tune to each repository. According to the company, most reviews now finish about 3x faster than preview, with typical review time going from about 20 minutes to 6-10 minutes. Anthropic said results now indicate the problem, what triggers it, and the solution in a few sentences, and administrators can assign each repository a review level of Rapid, Standard, or Deep so that routine code bases receive a quick pass while more critical ones receive a thorough review.

The company said preview users had said the system was catching real problems before a human reviewer reviewed a pull request, but that they wanted quieter reviews, adapted to each change, and faster. Anthropic also runs Code Reviews on its pull requests, which it says shows where reviews slow down its engineers.

To enable Code Review, an administrator turns it on in Claude Code settings, installs the Claude GitHub app, and selects repositories. Once enabled, reviews are automatically performed on new pull requests, with no configuration required by developers.

How code reviews work

According to Anthropic’s Code Review documentation, each review is performed on the Anthropic infrastructure in four phases: several agents examine the gap and the surrounding code in parallel to find issues, a verification phase compares candidate results against the code and filters out false positives, the remaining results are deduplicated and classified by severity, and the results are published in the pull request.

Each review leaves inline comments on the affected lines, a summary, and a review loop called Claude Code Review that lists the results. The check run never ends in failure, and a review never approves or blocks the pull request, so the merge decision remains with the team. When a review finds nothing, running the check reports that no issues were found, and Claude posts the same note on the pull request.

By default, Code Review focuses on correctness, that is, bugs that could break production rather than formatting preferences or test coverage failure. Repositories can optimize what is tagged with the REVIEW.md and CLAUDE.md help files, and the documentation states that REVIEW.md is read from the base or default branch, never from pull request changes. Code Review selects which Claude models reviews are performed on, the admin settings offer no model choice, and a deeper level of review costs more.

Each repository can perform reviews once when opening a pull request, after each push, or only on request, and anyone with write access can start one by commenting @claude review. Drafts and forks are excluded from automatic reviews. Reviews run in a managed environment with read-only repository access, work by reading diffs and surrounding code without compiling or running it, and do not run in self-hosted environments.

The service requires a Team or Enterprise plan and repositories hosted on GitHub, including GitHub Enterprise Server and GitHub Enterprise Cloud with data residency. It is not available to organizations with Zero Data Retention enabled or with HIPAA configuration enforced, it is not covered by the Anthropic Business Associate Agreement, and it is not yet available to organizations using customer-managed encryption keys. A Claude organization is also required when developers use Claude Code through Amazon Bedrock, Google Cloud Agent Platform, or Microsoft Foundry. The documentation warns that Code Review uses artificial intelligence and may not detect bugs, security vulnerabilities, or other issues, and that a review with no results does not mean the code is error-free or safe.

Search preview and reported internal results

Anthropic introduced Code Review as a search preview for Team and Enterprise plans on March 9, 2026, describing a system that sends a team of agents to every pull request and runs on nearly every internal pull request. The company said code output per engineer grew 200% over the previous year and that code review had become a bottleneck. Before Code Review, Anthropic reported, 16% of its pull requests received substantive review comments, compared to 54% after implementation.

On internal pull requests with more than 1,000 lines changed, Anthropic said 84% received results with an average of 7.5 issues, while pull requests of fewer than 50 lines fell to 31% with an average of 0.5 issues, and engineers marked less than 1% of the results as bad. The company also said that using early access exposed a latent type mismatch bug in the open source TrueNAS middleware that silently cleared the ZFS encryption key cache on every sync, a pre-existing issue in adjacent code touched by a pull request.

During preview, reviews were billed based on token usage and typically averaged $15-25 per review, based on the size and complexity of the pull request, and administrators had monthly organizational limits, repository-level controls, and an analytics dashboard that tracked reviewed pull requests, acceptance rate, and review costs.

Promotional credits for corporate organizations

Claude Enterprise organizations eligible to pay can claim between $10,000 and $50,000 in Code Review credits, with the amount determined at the time of claim based on the organization’s active Claude Code seats, under the terms of the Claude Help Center offer. An Owner or Primary Owner makes a request from the Code Review section of your organization’s settings; the request cannot be canceled or repeated and begins a 30-day period in which reviews are paid from credits first, and ends when the 30 days pass or the amount is depleted, whichever comes first. The request window opened on October 15, 2026 and closes at 11:59 PM UTC on January 15, 2027, and the help center recommends setting up Code Reviews on repositories before requesting, as the 30 days start from the time of complaint regardless of whether reviews are in progress or not.

The offer excludes organizations with reseller billing, organizations that have connected a GitHub Enterprise Server or data residency instance, expired trials and subscriptions, Team, Pro, Max, and free plans, organizations that have claimed Anthropic’s previous free Code Review offer, and organizations that cannot use Code Review, such as those with zero data retention. The actual cost of each review completed at list prices is derived first from credits; Reviews that fail or are stopped due to an issue on Anthropic’s part are not billed and do not count, while reviews that are stopped by a new push or organization count towards the cost until they are stopped. As long as credits remain, reviews can be performed even with a $0 spending limit. Credits do not cover self-hosted GitHub action, code review on GitHub Enterprise Server, /code-review OR /ultrareview commands or any other use of Claude. Organizations that do not submit requests by the January 15, 2027 deadline do not receive credit and reviews are billed as normal.