CodeRabbit (2026 Edition — CodeRabbit 2.x)

English | Bahasa Indonesia

---

<a name="english"></a>

English

Orchestration & Integration

• Anti-Slop & Autonomous Review: anti-slop, ai-code-review-autonomous, zero-tech-debt-auditor
• TDD & Testing Automation: autonomous-tdd-debugger, e2e-testing-expert, property-mutation-testing-expert
• CI/CD Quality Gates: ci-cd-devops-architect, github-actions-expert
• Architecture & Linters: scalability-clean-code, biome-linter-formatter-expert
• Lifecycle & Governance: brainstorming, zero-to-prod-orchestrator, session-memory-manager

Description

Expert guide for configuring and leveraging CodeRabbit 2.x — AI-powered automated code review, pull request summarization, and interactive developer feedback. Covers configuration, custom review rules, integration with CI/CD, and agentic review mode.

Trigger Conditions

• Setting up CodeRabbit on a GitHub or GitLab repository.
• Configuring custom review rules for your tech stack.
• Using CodeRabbit's agentic mode for deep architectural reviews.
• Integrating CodeRabbit feedback into CI/CD quality gates.
• Writing CodeRabbit-compatible custom instructions in .coderabbit.yaml.

CodeRabbit 2.x Key Features

Feature	Description
PR Summary	Auto-generates structured PR description with walkthrough
Line-by-line Review	Inline comments with specific fix suggestions
Agentic Review	Deep analysis spanning multiple files for architectural issues
Learnable Rules	Learns from review dismissals and approvals over time
CI Integration	Blocks PR merge if critical issues found
Chat Mode	Ask CodeRabbit questions about the PR in review thread

Configuration (.coderabbit.yaml)

# .coderabbit.yaml — place in repository root
version: "2"
language: "en-US"
tone_instructions: "Be concise and direct. Prioritize security and performance issues."

reviews:
  auto_review:
    enabled: true
    drafts: false           # Don't review draft PRs
    base_branches: ["main", "develop"]

  profile: "chill"          # assertive | chill | default
  request_changes_workflow: false
  high_level_summary: true
  commit_message_instructions: "Use Conventional Commits format: feat/fix/chore/docs"

  # Ignore paths from review
  path_filters:
    - "!**/*.lock"
    - "!**/migrations/**"
    - "!**/generated/**"
    - "!**/__snapshots__/**"

  # Stack-specific review instructions
  path_instructions:
    - path: "apps/api/**"
      instructions: |
        - Check for missing input validation (Zod schemas required on all handlers)
        - Flag any database queries without proper RLS consideration
        - Ensure all async functions have try/catch or error boundaries
        - Check for N+1 query patterns (missing .include() or DataLoader usage)
    - path: "apps/web/**"
      instructions: |
        - Check for missing `key` props in .map() renders
        - Flag `useEffect` without proper cleanup functions
        - Check for missing `alt` attributes on images
        - Verify Server Actions are properly validated with Zod
    - path: "apps/admin/**"
      instructions: |
        - All admin routes must verify isSuperAdmin === true
        - Flag any direct DB access without service role client
        - Check for proper audit logging on destructive operations

  # Custom review rules for the entire codebase
  instructions: |
    Review with these priorities:
    1. SECURITY: SQL injection, XSS, auth bypass, exposed secrets
    2. CORRECTNESS: Logic bugs, off-by-one errors, type unsafety
    3. PERFORMANCE: N+1 queries, unnecessary re-renders, bundle size
    4. MAINTAINABILITY: Code duplication, naming, SOLID violations
    5. STYLE: Only comment if it's a significant clarity issue
    6. SOVEREIGN ANTI-SLOP: Flag placeholders (// TODO, // ...), mock data in prod, syntax comments, and silent error suppression

# PR summary format
summary:
  auto_title_placeholder: "🤖 AI Title"
  description: |-
    ## Summary
    <!-- Concise description of what changed -->
    
    ## Changes
    <!-- Structured list by area -->
    
    ## Testing
    <!-- What was tested -->

# Enable chat for interactive Q&A
chat:
  auto_reply: true

Agentic Review Mode (CodeRabbit 2.x)

CodeRabbit 2.x introduces agentic review — deep analysis that can read multiple files, run tools, and understand architectural context:

# Trigger agentic review in PR comment:
@coderabbitai review

# Ask specific questions:
@coderabbitai What is the security impact of changes in apps/api/src/routes/users.ts?
@coderabbitai Can you generate a test for the createWorkspace function?
@coderabbitai Is there a N+1 query issue in this PR?
@coderabbitai Summarize the architecture changes in this PR

CI/CD Integration (GitHub Actions)

# .github/workflows/pr-quality.yml
name: PR Quality Gate

on: [pull_request]

jobs:
  coderabbit-review:
    runs-on: ubuntu-latest
    steps:
      - name: Wait for CodeRabbit Review
        uses: coderabbit-ai/wait-for-review@v1
        with:
          github-token: ${{ secrets.GITHUB_TOKEN }}
          min-approvals: 1
          block-on-critical: true  # Block merge if critical issues found

CodeRabbit vs Manual Review — Best Practices

• CodeRabbit first: Let it run before human reviewers — catches the obvious issues.
• Human focus: After CodeRabbit, humans focus on business logic, architecture intent, and product decisions.
• Teach it: Dismiss false positives with an explanation — CodeRabbit learns your patterns.
• Custom instructions: Invest time in path_instructions for your specific stack — dramatically improves relevance.

---

<a name="bahasa-indonesia"></a>

Bahasa Indonesia

Integrasi Orkestrasi

• Anti-Slop & Review Otonom: anti-slop, ai-code-review-autonomous, zero-tech-debt-auditor
• TDD & Otomasi Pengujian: autonomous-tdd-debugger, e2e-testing-expert, property-mutation-testing-expert
• Gerbang Kualitas CI/CD: ci-cd-devops-architect, github-actions-expert
• Arsitektur & Linter: scalability-clean-code, biome-linter-formatter-expert
• Siklus Hidup & Tata Kelola: brainstorming, zero-to-prod-orchestrator, session-memory-manager

Deskripsi

Panduan ahli untuk mengonfigurasi dan memanfaatkan CodeRabbit 2.x — review kode otomatis bertenaga AI, ringkasan pull request, dan umpan balik developer interaktif. Mencakup konfigurasi, aturan review kustom, integrasi dengan CI/CD, dan mode review agentik.

Kondisi Pemicu

• Menyiapkan CodeRabbit di repositori GitHub atau GitLab.
• Mengonfigurasi aturan review kustom untuk tech stack Anda.
• Menggunakan mode agentik CodeRabbit untuk review arsitektur mendalam.
• Mengintegrasikan umpan balik CodeRabbit ke dalam quality gate CI/CD.
• Menulis instruksi kustom yang kompatibel dengan CodeRabbit di .coderabbit.yaml.

Fitur Utama CodeRabbit 2.x

Fitur	Deskripsi
Ringkasan PR	Membuat deskripsi PR terstruktur otomatis dengan walkthrough
Review Baris per Baris	Komentar inline dengan saran perbaikan spesifik
Review Agentik	Analisis mendalam yang mencakup banyak file untuk masalah arsitektur
Aturan yang Dapat Dipelajari	Belajar dari penolakan dan persetujuan review dari waktu ke waktu
Integrasi CI	Memblokir merge PR jika masalah kritis ditemukan
Mode Chat	Tanyakan pertanyaan kepada CodeRabbit di thread review

Konfigurasi (.coderabbit.yaml)

Tempatkan .coderabbit.yaml di root repositori. Konfigurasi kunci:

• reviews.profile: assertive (ketat), chill (santai), default.
• reviews.path_instructions: Instruksi review kustom per direktori — sangat berguna untuk menerapkan standar spesifik stack.
• reviews.path_filters: Kecualikan file lock, migrasi, kode yang dihasilkan dari review.
• reviews.instructions: Aturan global untuk seluruh codebase dengan prioritas eksplisit.

Mode Review Agentik (CodeRabbit 2.x)

Mode agentik memungkinkan CodeRabbit membaca banyak file, menjalankan tool, dan memahami konteks arsitektur. Picu dengan @coderabbitai review di komentar PR atau ajukan pertanyaan spesifik tentang dampak keamanan, masalah N+1, atau perubahan arsitektur.

Integrasi CI/CD (GitHub Actions)

Gunakan action coderabbit-ai/wait-for-review@v1 untuk menunggu review selesai dan memblokir merge jika masalah kritis ditemukan.

CodeRabbit vs Review Manual — Best Practices

• CodeRabbit dulu: Biarkan berjalan sebelum reviewer manusia — menangkap masalah yang jelas.
• Fokus manusia: Setelah CodeRabbit, manusia fokus pada logika bisnis, niat arsitektur, dan keputusan produk.
• Ajarkan: Tolak false positive dengan penjelasan — CodeRabbit belajar pola Anda.
• Instruksi kustom: Investasikan waktu di path_instructions untuk stack spesifik Anda — meningkatkan relevansi secara dramatis.