Claude Code Review is out of preview for Team and Enterprise
Admins can turn on PR reviews that Anthropic says mostly finish in 6 to 10 minutes and pick Quick, Standard or Deep per repo. Eligible Enterprise orgs can claim $10,000 to $50,000 in credits.
If your company is on Claude Team or Enterprise and keeps code on GitHub, an Owner can now turn on Code Review and have Claude review new pull requests automatically. Anthropic says "No configuration needed." Anthropic moved it out of research preview on Oct 15. Enterprise orgs that qualify can claim between $10,000 and $50,000 in review credits through Jan 15, 2027. Per the Help Center, "Your credits last 30 days from the moment you claim." Our advice: set up reviews before you claim.
What changed#
Anthropic's launch post says Code Review "is out of research preview and available to Claude Team and Enterprise organizations." Three things are new at general availability, per the post:
- Speed: "Most reviews now finish about 3x faster (~20 minutes down to 6-10 minutes) compared to during preview."
- Quality: findings that "state the problem, what triggers it, and the fix in a few sentences."
- Review levels: admins pick Quick, Standard (the default) or Deep for each repository. The docs say a more thorough level "costs more."
How it works hasn't changed. Several agents read the diff and the code around it, a verification step filters out false positives, and the rest land as inline comments marked Important, Nit or Pre-existing, plus a check run named Claude Code Review. A review "never approves or blocks the pull request," and the check run doesn't fail.
| Research preview (Mar 9) | General availability (Oct 15) | |
|---|---|---|
| Typical review time | "around 20 minutes" | "6-10 minutes" for most reviews |
| Depth control | none described | Quick, Standard, Deep per repo |
| Cost | $15 to $25 per review on average, per Anthropic | no flat price; billed on tokens (no new average given) |
Why it matters#
Anthropic's case is that coding agents keep raising the pace of new code, "and the review bottleneck hasn't eased." It says that on its own pull requests, the share getting substantive review comments went from 16% to 54% after it started using Code Review. That is Anthropic's internal number, not an independent test, and the docs warn that "a review with no findings doesn't mean your code is error-free or secure." For an engineering lead, the real decisions are cost and noise. You control cost with the review level and how often reviews run, and you control noise with a REVIEW.md file in the repo.
How to use it#
For the Owner of your Claude organization (from the setup and costs docs):
- Go to Admin settings > Claude Code on claude.ai, find Code review, and click Allow code review.
- Connect GitHub and install the Claude GitHub App on the repositories you want. You need to be an owner of the GitHub organization, or have one install it.
- Set a monthly spend limit for the Claude Code Review service at claude.ai/admin-settings/usage before a wide rollout.
- Pilot a few repos with Review runs set to Once after PR creation or Manual (@claude review). New repos start at After every push, which runs the most reviews.
- Open a non-draft test PR. A Claude Code Review check run should appear within a few minutes. (In a Manual repo, comment
@claude reviewto start it.) - After several reviews, read the Average cost column in the Repositories table (per the costs docs). Our estimate method: multiply that average by the PRs the repo opens in a month, and by pushes per PR if you use After every push.
Pilot cost estimate (ours). Anthropic has published no per-review price for general availability; reviews are "billed on token usage." The only published figure is the preview average, which Anthropic put at $15 to $25 per review. Using that range, a repo opening 100 PRs a month costs about 100 x $15 = $1,500 to 100 x $25 = $2,500 a month on Once after PR creation. At 3 pushes per PR on After every push, that is 300 reviews, or $4,500 to $7,500. Replace the $15 to $25 with your own Average cost once reviews run. 7. Enterprise only: once reviews are running, claim the credits under Organization settings > Claude Code > Code Review.
For developers: comment @claude review for one review, or @claude review always to review every push on that PR. You need write access to the repository.
To cut nits and skip generated files, commit a REVIEW.md at the repo root and merge it. This starter is ours, not Anthropic's:
# Review instructions (our starter)
## Severity
Mark Important only for bugs that break behavior, leak data, or block a rollback.
Style and naming suggestions are Nit at most.
## Limits
Report at most three Nits per review.
## Skip
- Anything our CI already checks: lint, formatting, type errors
- Generated code under gen/ and any lockfile
## Always check
- New API endpoints have a test
- Logs never include email addresses or request bodies
Good to know#
- Credits fine print. The amount is set at claim time, "between $10,000 and $50,000, based on your organization's active Claude Code seats." You "can claim only once." Not eligible: Team plans, trials, orgs billed through a reseller, orgs that claimed the earlier free offer, and orgs with GitHub Enterprise Server or a
.ghe.comdata residency instance. Connect one of those after claiming and the credits "stop applying for the whole organization." - Who can't use it at all. It isn't available with Zero Data Retention or the HIPAA configuration, isn't covered by Anthropic's BAA, and isn't available yet with customer-managed encryption keys. You need a Team or Enterprise Claude org even if your developers use Claude Code through Bedrock, Google Cloud or Foundry.
- Billing. Reviews are billed to the organization in usage credits, separate from plan usage. When a spend limit or balance runs out, reviews are skipped and the PR gets a comment saying why.
REVIEW.mdmust be merged. It's read from the base or default branch, "never from the pull request's own changes." The docs say instructions that lower severity, skip paths or cap nits are honored from the root file when it sits on a protected default branch, so protect that branch.- Drafts and forks are skipped by automatic reviews; use
@claude reviewon them. - Clash with the GitHub Action. If a repo also runs the Claude Code GitHub Action with its default
@claudetrigger,@claude reviewstarts the Action too. Change the Action'strigger_phrase. - No model choice. Code Review picks the models. The docs say that inside a review, "access to your repository is read-only, and a review doesn't push commits."
- The local option is different.
/code-reviewin a Claude Code session works on any plan but readsCLAUDE.md, notREVIEW.md, and the credits don't cover it.
What to watch: whether Anthropic publishes typical per-review costs for each level now that it's out of preview, and whether the credit offer extends to GitHub Enterprise Server and data residency orgs.
Sources#
- Code Review is now generally available (Anthropic, Oct 15)
- Code Review docs, setup, costs, customize, pull requests
- Enterprise promotional credits (Claude Help Center)
- Bringing Code Review to Claude Code (research preview post, Mar 9)
- How the Claude Code team does code review (claude.dev, Oct 15)
- @ClaudeDevs launch thread (Oct 15, 6:51 PM ET)
Last checked: 2026-10-16