Claude Code Review is generally available: what to know
Claude Code Review is out of research preview for Team and Enterprise: what it checks, how an Owner turns it on, what it costs and where it falls short.
Claude Code Review is a managed service that reviews the pull requests in your GitHub repositories, and it is now out of research preview for Claude Team and Enterprise organizations. It posts bugs as inline comments, never approves or blocks a merge, and is billed on token usage through usage credits. An Owner of your Claude organization turns it on.
This site is independent and not affiliated with Anthropic. We read Anthropic's announcement and its Code Review docs on 2026-10-16. We did not turn the service on or run a review, so nothing below is a hands-on result.
What a review does
Several agents read each change next to the code around it, looking for logic errors, security holes and broken edge cases. A verification step then checks each finding against the code before it is reported. By default the focus is correctness: "bugs that would break production, not formatting preferences or missing test coverage."
Each review leaves findings as inline comments, a summary, and a check run named Claude Code Review. Anthropic's docs are firm about the limit: a review never approves or blocks the pull request, so the decision to merge stays with your team.
What changed at general availability
The announcement says Code Review "is out of research preview and available to Claude Team and Enterprise organizations." Three details matter:
- Speed. "Most reviews now finish about 3x faster (~20 minutes down to 6-10 minutes) compared to during preview."
- Levels. "Admins choose Quick, Standard (default), or Deep for each repository, so routine codebases get a quick pass and the most critical ones get a deep review."
- Credits. "Eligible Claude Enterprise organizations can claim at least $10,000 in Code Review credits by January 15, 2027. Credits expire 30 days after you claim them."
The announcement makes its case in one line: "Coding agents keep increasing the pace at which teams write code, and the review bottleneck hasn't eased."
Reading what it posts
Every finding carries one of three markers, per the pull request docs:
| Marker | Severity | Meaning |
|---|---|---|
| 🔴 | Important | A bug that should be fixed before merging |
| 🟡 | Nit | A minor issue, worth fixing but not blocking |
| 🟣 | Pre-existing | A bug that exists in the codebase but wasn't introduced by this PR |
To clear a finding, push a fix or resolve its thread on GitHub. Replying alone does not dismiss it. The next review reads your replies and can resolve the thread or post one follow-up.
Who turns it on, and the first test
Only an Owner can do it, in Admin settings > Claude Code on claude.ai. The setup page lists the steps. Reviews run on Anthropic infrastructure, so you add no workflow files to your repositories.
Click Allow code review
In the Code review section of Admin settings > Claude Code.
Connect GitHub and install the Claude GitHub App
The GitHub organization owner may need to do this part.
Choose repositories
You need write access to each one you add.
Change Review runs before you rely on it
New repositories start at After every push, the setting that runs the most reviews.
Open a test pull request
Not a draft. A check run named Claude Code Review should appear within a few minutes.
If you manage the organization, do step four first. Anthropic's costs page recommends a small pilot on a few repositories set to Once after PR creation or Manual (@claude review).
The Claude moves worth stealing, by email
Workflows tested on real jobs, sent only when they are worth it. Free.
What it costs
Anthropic's costs page says Code Review "has no flat per-review price." It is billed on token usage through usage credits, separately from your plan's included usage, and billed to the organization, not to the developer who opened the pull request.
What moves the bill:
- Review level. A more thorough review costs more.
- Size of the change. Bigger pull requests cost more to review.
- How often reviews run. After every push multiplies the cost by the number of pushes.
To cap it, set a monthly spend limit for the Claude Code Review service at claude.ai/admin-settings/usage. When a limit runs out, reviews are skipped and the pull request gets a comment saying why. For the cost of Claude Code itself, see our Claude Code pricing guide.
Tell it what to flag
You change what Code Review flags by committing a REVIEW.md file at your repository root. Anthropic's customization page has a full example. This is a shortened piece of it, which a team tired of lint comments could start from:
# Review instructions
## Cap the nits
Report at most five Nits per review.
## Do not report
- Anything CI already enforces: lint, formatting, type errors
- Generated files under `src/gen/` and any `*.lock` file
Merge the file before you expect any effect. Code Review reads REVIEW.md from the pull request's base branch or the default branch, "never from the pull request's own changes." Anthropic recommends protecting the default branch, because some instructions, such as lowering a severity, excluding paths or capping nits, need the root file on a protected default branch.
Project rules you want everywhere belong in CLAUDE.md, which the service also reads. Our guide on sharing AGENTS.md and CLAUDE.md covers that file.
No Team or Enterprise plan?
The managed service is for Team and Enterprise. The /code-review command is not. Anthropic's local review page says it "works on any plan" and runs in your own session, so it counts toward your plan's usage or API billing like the rest of the session.
Run /code-review for your current branch, or point it at a target such as /code-review main...my-feature. Add --fix to have Claude apply the findings, subject to your permission mode. It runs as a background subagent, so it does not fill your conversation.
One difference to remember: the command follows CLAUDE.md but does not read REVIEW.md. For proving a change works rather than reading it, see our verification guide.
Where it falls short
Anthropic's own note: "Code Review uses AI to review code and may miss bugs, security vulnerabilities, or other issues. A review with no findings doesn't mean your code is error-free or secure." Treat it as an extra reader, not a gate.
- It reads, it does not run. A review works from "reading the diff and the surrounding code, without building or running your code."
- GitHub only. The managed service covers GitHub repositories, including GitHub Enterprise Server and Cloud with data residency.
- Some organizations are excluded. It is not available with Zero Data Retention or the HIPAA configuration, and not yet with customer-managed encryption keys.
- No self-hosted environments. Reviews do not run in them.
- Cost surprises. A repository left at After every push on an active project can run many reviews per pull request.
Does Claude Code Review approve or block pull requests?
No. Anthropic's docs say a review never approves or blocks the pull request, and the check run never concludes as a failure. The merge decision stays with your team, unless your own branch protection requires findings to be resolved first.
Can I use Code Review on Pro or Max?
The managed service is for Team and Enterprise organizations. On any plan you can run /code-review inside Claude Code instead. It reviews your branch locally and counts toward your plan's usage or API billing like the rest of the session.
How do I get a review on a draft or fork pull request?
Comment @claude review as a new top-level comment. Automatic reviews skip drafts and forks. You need write access to the repository, and for a fork that means the base repository. Bots cannot start reviews this way.