AI Code Review

Code reviews that think like your team.

An AI reviewer that learns your architecture, your standards, your opinions. Not generic best practices — yours.

reviewbot
$ reviewbot reviewing PR #847...
 
⚠ src/modules/accounts/services/accounts.service.ts:42
│ This imports directly from policies/repository — violates
│ your module boundary rule (ARCH-001). Use PoliciesService
│ through DI instead.
│ Why: Your team is preparing for microservice extraction.
│ Every cross-module repo import is an extraction blocker.
 
✓ 2 issues found · 0 noise · reviewed in 8s
How it works
01CONNECT

Connect your GitHub or GitLab repo. The bot analyzes your codebase structure, patterns, and conventions.

$ reviewbot connect
Scanning repository...
Found 847 files across 23 modules
Detected: TypeScript, React, NestJS
Architecture: modular monolith
> Ready for interview.
02INTERVIEW

The bot interviews your team in a 15-minute chat. It learns your rules, your style, your opinions.

Q: How should services communicate?
[x] Event bus (Kafka/RabbitMQ)
[ ] Direct imports
[ ] HTTP calls between services
[ ] Shared database
> Noted: "strict module boundaries"
03REVIEW

Every PR gets reviewed like your best engineer would. No noise. No generic advice. Just what matters.

$ reviewbot review PR #291
accounts.service.ts:42
Cross-module internal import.
Use PoliciesService through DI.
1 issue found · 0 noise · 4s
Why ReviewBot

Generic bots add noise. We add signal.

Other bots
ReviewBot
ERROR_HANDLING

"Consider adding error handling"

ERROR_HANDLING

"This handler silently swallows TransactionFailed. Your team requires explicit error propagation per ADR-012."

ARCHITECTURE

"Consider breaking this into smaller functions"

ARCHITECTURE

"This bypasses your CQRS pattern. Writes in RPC handlers should go through the event bus per your architecture decision."

CODE_STYLE

"Use camelCase for variable names"

CODE_STYLE

"Your monorepo uses snake_case in the data layer (see /packages/db). Only API route handlers use camelCase."

TESTING

"Add unit tests"

TESTING

"This payment flow needs integration tests with Stripe fixtures, matching your /tests/integration pattern."

LEARNS

Your codebase, your rules, your opinions

ZERO CONFIG

15-min interview replaces hours of rule writing

NO NOISE

Only flags what your team actually cares about

CONTEXT-AWARE

Understands architecture, not just syntax

Pricing
Free
$0/month

For solo devs and small projects.

  • 1 repository
  • 50 reviews/month
  • Predefined skills only
  • Bring your own key
Get Started
Pro
$20/month

For teams that ship fast.

  • Unlimited repositories
  • Unlimited reviews
  • Custom skills
  • Bring your own key
Start Pro

Bring your own API key. You pay only for LLM tokens.

Both plans use your own API key. Average cost per review is ~$0.10.

150 PRs/month
= ~$15/month in LLM tokens