A cast, not just a bot.
Fourteen reviewers read every PR.

Re:View fields named specialists — Grace on correctness, Bruce on security, Martin on consistency with the code you already have, and eleven more. Grace reads every pull request and calls in the specialists a change needs.

Free for private repos to start · full cast on open source · no card required

The cast

Fourteen specialists. One review per persona, every time.

Every persona is a focused reviewer with its own specialty — a distinct failure mode it alone catches. Enable the cast your codebase needs, or summon one on a PR by commenting their specialty, like @re-view security or @re-view refactor. @re-view all calls the full cast.

Core cast — every paid plan

Grace

General review

Reads everything, judges like a senior generalist. Correctness, behavior risk, contract hygiene.

@re-viewgeneral

Bruce

Security

Injection, authz gaps, leaked secrets, weak crypto. Assumes the diff is hostile until proven safe.

@re-viewsecurity

Frank

Architecture

Coupling, boundaries, placement, API surface. How it is built outlasts what it does.

@re-viewarchitecture

Maya

Documentation

Doc drift, missing reference for new public surface, stale examples. An undocumented change is half a change.

@re-viewdocs

Kent

Tests

Untested paths, weak assertions, flaky patterns. If it is not tested, it is broken.

@re-viewtests

Martin

Consistency & refactoring

Searches your repo before he comments: helpers you already have, logic now kept in two places, conventions the change breaks, dead code left behind. New code should look like it always lived here.

@re-viewconsistencyrefactor

Specialists — Studio plan

Leslie

Concurrency & distributed

Races, deadlocks, idempotency, partial failure. Time and ordering are her home turf.

@re-viewconcurrencydistributed

Brendan

Performance

N+1s, accidental quadratic, allocation churn. Finds the hot path before the pager does.

@re-viewperf

Jony

UI/UX & design systems

Design-token drift, missing states, interaction hygiene. The details are the product.

@re-viewuiux

Judy

Accessibility

WCAG 2.2 essentials: semantics, keyboard paths, screen-reader experience, contrast.

@re-viewa11y

Edgar

Data & schema

Migrations, constraints, indexes, query shape. Relationships done right.

@re-viewdataschema

Cleo

i18n & localization

Hardcoded strings, broken plurals, RTL, encoding. Every user, their language.

@re-viewi18nl10n

Betsy

Reliability & observability

Swallowed errors, missing timeouts, log hygiene. You cannot operate what you cannot see.

@re-viewreliability

Larry

Licensing & privacy

Incompatible licenses, PII flows, data sinks. What you may copy, what you must protect.

@re-viewlicenseprivacy

How it works

Running in under five minutes.

01

Install the App

One click from the GitHub marketplace. Pick the repos. No CI config, no workflow files, nothing to maintain.

02

Choose your cast

A small YAML in the repo default branch — or the dashboard. Trigger mode, personas, path filters, review language.

03

Reviews land in the PR

Grace reviews each pull request and calls in specialists when the change needs them; @re-view security brings in Bruce on demand.

# .re-view.yml (default branch) reviews: trigger: every_push # or opened | requested personas: [grace, bruce, kent] paths: { exclude: ["**/*.md", "vendor/**"] } language: en

Built for trust

Reviews only — nothing lands without your approval.

Reviews only

Re:View reads your code and posts findings. No branches, no commits, no merges.

Sandboxed by default

Every review runs in an isolated, egress-locked container. Your code cannot phone home.

You pick when reviews run

When a pull request opens, on every push, or only when asked. Paid plans get every mode; Free reviews on open or on request.

Predictable bills

Flat per-seat pricing with unlimited reviews under fair use. No per-file meters, no overage, no surprises.

Talk to a reviewer

@re-view review, @re-view security, @re-view refactor — summon exactly the review you need, in the PR.

Reviews in your language

One setting, and the cast writes findings in your team's language.

Pricing

Simple seats, unlimited reviews.

Pay per seat. Paid plans review every pull request your team opens, under fair use. Free covers 25 reviews a month.

Free

$0

  • Grace on private repos
  • 25 reviews / month
  • Review on open or on request
  • Full cast on open source
Start free

Core

$19/user/mo

  • Core cast: Grace, Bruce, Frank, Maya, Kent, Martin
  • Unlimited reviews (fair use)
  • Dashboards + usage meters
Start 14-day trial

Enterprise

Custom

  • SSO/SAML + SCIM, audit logs
  • VPC or self-hosted runners
  • Custom personas + policy packs
  • SLA, data residency, invoicing
Talk to us

Annual billing saves ~20%. Open-source repositories get the full cast free under fair-use caps — because public code deserves public-interest review. Paid plans have no overage charges. Prices exclude tax. Every new paid subscription comes with a 30-day money-back guarantee.

FAQ

Questions engineers ask.

Can Re:View touch my code?

No. Re:View can read repository contents, and it has read and write access to pull requests and issues so it can post reviews and comments, and it publishes a Re:View approval check that your branch rules can require. It creates no branches, makes no commits, and merges nothing. On small documentation-only pull requests it may post an approving review instead of running the full cast; your branch protection rules decide whether that approval counts. Reviews run in single-use sandboxed containers with network access limited to GitHub's API and our AI model providers.

What models power the reviewers?

Reviews are written by Anthropic Claude models, served by Anthropic and reached through OpenRouter. A small Google model first picks which specialists a pull request needs, from the list of changed file paths only. You pay per seat, never per token, and we do not use your code to train models. See Security for what is sent and kept.

What does "unlimited reviews (fair use)" mean?

Core and Studio have no review quota and no overage charges: every pull request your team opens gets reviewed. Fair use covers normal team development. If automated traffic pushes an account far beyond typical volume, we'll contact you to talk it through. We won't cut reviews off or bill per review. Free includes 25 reviews a month, and open-source repositories run the full cast under a separate fair-use budget.

When do reviews happen?

Your choice, for the whole organization in the dashboard or per repository in .re-view.yml: opened reviews each new pull request, every_push reviews again on every commit, and requested waits until someone adds Re:View as a reviewer or comments @re-view. Requests always work, whatever the setting.

How do I turn a reviewer off for a repo?

Set the personas list in .re-view.yml on the default branch, or toggle it in the dashboard. Config can only narrow — a pull request can never mute its own review or unlock a persona above the plan.

Is my code used for training?

No. We don't use your code to train models, and we only send it to Anthropic's own API, through OpenRouter, under terms that prohibit training on it. Providers may keep requests briefly for abuse and safety monitoring under their API terms. We don't keep copies of your code after the review.

Can I get a refund?

Yes, within 30 days. If a new Core or Studio subscription isn't working for you, email contact@re-view.ai within 30 days of its first payment, monthly or annual, and we'll refund that payment in full. After that you can cancel any time and keep access until the end of the period you've paid for; other payments aren't refunded except where the law requires. Payments are processed by Stripe through Link, which calculates any sales tax or VAT at checkout. Details are in our Terms.

What about self-hosting?

Enterprise plans can run review workers inside your own VPC. Talk to us about data-residency and compliance requirements.