Anthropic announced on October 15, 2026, that Claude Code Review had left research preview and was generally available to Claude Team and Enterprise organizations. The managed service reviews GitHub pull requests and posts findings, while the engineering team retains the merge decision.
Anthropic says most reviews now take about 6–10 minutes, compared with approximately 20 minutes during the preview. The company also introduced three review-depth settings: Quick, Standard and Deep. (Anthropic’s announcement)
How Claude Code Review examines a pull request
The service uses parallel agents to look for potential issues, verifies candidate findings, removes duplicates and ranks results by severity before posting them. Findings appear as inline comments, alongside a summary and a check run. By default, reviews focus on correctness issues with possible production impact, rather than formatting preferences or missing test coverage.
Administrators enable the service, install Claude’s GitHub App and choose repositories. They can set Quick, Standard—the default—or Deep review for each repository; deeper reviews cost more. Repositories can run reviews when a pull request opens, after each push or only on request. A user with write access can also comment @claude review. Draft pull requests and forks are excluded from automatic reviews. Teams can supply repository guidance in REVIEW.md and CLAUDE.md; REVIEW.md is read from the base or default branch.
Anthropic’s product documentation says the managed service has read-only repository access and does not build or run the code. It does not approve or block pull requests, so the team decides whether findings warrant changes and when to merge. Anthropic also warns that a review with no findings does not establish that code is bug-free or secure.
Who can use the managed service
Claude Code Review is available to Claude Team and Enterprise organizations with GitHub repositories. Anthropic’s documentation covers GitHub Enterprise Server and GitHub Enterprise Cloud with data residency, but the service is unavailable to organizations configured for Zero Data Retention or HIPAA, and to those using customer-managed encryption keys. It is not covered by Anthropic’s Business Associate Agreement. A Claude organization is required even when Claude Code is accessed through Amazon Bedrock, Google Cloud Agent Platform or Microsoft Foundry.
How the Enterprise credit offer works
Eligible paid Claude Enterprise organizations can claim $10,000–$50,000 in US-dollar credits for qualifying Code Review activity. The amount is set once, based on active Claude Code seats. An Owner or Primary Owner can make one claim in the organization’s Code Review settings from October 15, 2026, through January 15, 2027, at 23:59 UTC. The 30-day credit period starts when the claim is made, and unused credits expire when it ends.
The credits apply first to qualifying reviews of repositories hosted at github.com. If credits do not cover a review’s full cost, the remainder is billed under the organization’s usual limits. Paying through a cloud marketplace does not count as reseller billing. However, connecting GitHub Enterprise Server or GitHub Enterprise Cloud with data residency after claiming stops credits across the organization, including for github.com reviews. The offer excludes the self-hosted GitHub Action, GHES reviews, /code-review, /ultrareview and other Claude use. (Anthropic Help Center’s offer terms)
