Intelligent code review automation for GitLab using Dify AI with RAG support for frontend best practices.

• Features
• Architecture
• Prerequisites
• Quick Start
• Installation
• Configuration
• Knowledge Base Setup
• Usage
• Project Structure
• Development
• Deployment
• Testing

• Quick Start Guide - Get started in 5 minutes
• Cost Analysis - Detailed pricing breakdown ($0.003-0.04/file)
• Knowledge Base Guide - RAG setup and maintenance
• CLI Documentation - Command-line interface reference
• FAQ - Frequently asked questions

• 🤖 AI-Powered Reviews: Leverages Dify API for intelligent code analysis
• 📚 RAG Integration: Context-aware reviews using best practices knowledge base
  • React, Vue, Angular best practices
  • Accessibility guidelines (WCAG)
  • Performance optimization patterns
  • Security best practices
• 🎯 Repository Context: Custom context files for project-specific guidelines
  • Monorepo/polyrepo structure awareness
  • Project-specific code style and standards
  • Custom skip patterns and focus areas
  • Technology stack considerations
• 🔄 GitLab Webhook: Automatic MR review triggers
• 🌐 Multi-Language: Supports 15+ programming languages
• 🔒 Security Analysis: Detects vulnerabilities and security issues
• 📊 Detailed Reports: Comprehensive review with actionable insights
• ⚡ Performance: Async processing with rate limiting
• 🔌 Local Fallback: Works even when Dify RAG is unavailable
• 🧪 Test Coverage: Built-in testing suite

┌─────────────────────────────────────────────────────────────┐ │ GitLab MR │ │ (Merge Request) │ └────────────────────┬────────────────────────────────────────┘ │ Webhook Event ▼ ┌─────────────────────────────────────────────────────────────┐ │ Webhook Handler │ │ (Validates & Dispatches) │ └────────────────────┬────────────────────────────────────────┘ │ ▼ ┌─────────────────────────────────────────────────────────────┐ │ Review Service │ │ (Orchestrates Review Flow) │ └──────┬──────────────────────┬───────────────────────────────┘ │ │ ▼ ▼ ┌──────────────┐ ┌──────────────────┐ │ GitLab │ │ RAG Service │ │ Service │ │ (Best Practices)│ └──────┬───────┘ └────────┬─────────┘ │ │ │ Get Diffs │ Query Knowledge ▼ ▼ ┌──────────────┐ ┌─────────────────┐ ┌──────────────┐ │ MR Changes │ │ Dify RAG API │ │ knowledge- │ │ (Diffs) │ │ (Primary) │ │ base/ │ └──────┬───────┘ └────────┬────────┘ │ (Fallback) │ │ │ └──────┬───────┘ │ │ If unavailable │ │ └───────────────────────┘ │ │ └────────────────┬───────────────┘ ▼ ┌───────────────┐ │ Dify Service │ │ (AI Review) │ └───────┬───────┘ │ ▼ ┌───────────────┐ │ Review Result │ └───────┬───────┘ │ ▼ ┌───────────────────┐ │ GitLab Comment │ │ (Posted to MR) │ └───────────────────┘ 

knowledge-base/ ├── frontend/ │ ├── react-best-practices.md ──┐ │ ├── angular-best-practices.md ─┤ │ ├── accessibility.md ──────────┼─→ Upload Script │ ├── performance.md ────────────┤ (npm run setup:knowledge-base) │ └── security.md ───────────────┘ │ ▼ ┌──────────────┐ │ Dify RAG │ │ Dataset │ └──────┬───────┘ │ ┌──────────────────────────┼────────────────────────┐ │ During Review │ │ ▼ ▼ ▼ ┌──────────┐ ┌─────────────┐ ┌──────────────┐ │ Analyze │ Query │ Retrieve │ Use │ Enhanced │ │ File │────────────│ Relevant │─────────│ AI Review │ │ Type │ │ Context │ │ with Best │ └──────────┘ └─────────────┘ │ Practices │ └──────────────┘ │ │ If Dify unavailable ▼ ┌──────────────┐ │ Local RAG │ │ Fallback │ │ (Read .md) │ └──────────────┘ 

• Node.js >= 22.x
• GitLab account with API access
• Dify account with API key
• Docker (optional, for containerized deployment)

# Clone repository git clone https://github.com/your-org/gitlab-ai-review.git cd gitlab-ai-review # Run automated setup npm run setup

The setup script will:

• ✅ Install dependencies
• ✅ Create .env from template
• ✅ Validate Node.js version
• ✅ Create logs directory
• ✅ Optionally upload knowledge base to Dify

# Install dependencies npm install # Copy environment template cp .env.example .env # Edit configuration nano .env

Create .env file:

# Server Configuration NODE_ENV=production PORT=3000 LOG_LEVEL=info # GitLab Configuration GITLAB_TOKEN=glpat-your_token_here GITLAB_URL=https://gitlab.com GITLAB_WEBHOOK_SECRET=your_webhook_secret # Dify Configuration DIFY_API_KEY=app-your_dify_key DIFY_API_URL=https://api.dify.ai/v1 DIFY_USER=gitlab-bot # Review Configuration MAX_FILES_PER_REVIEW=20 MAX_DIFF_SIZE=5000 ENABLE_INLINE_COMMENTS=false SKIP_PATTERNS=node_modules,dist,build,*.lock # Rate Limiting RATE_LIMIT_WINDOW_MS=900000 RATE_LIMIT_MAX_REQUESTS=100 # RAG Configuration RAG_KNOWLEDGE_BASE=frontend-best-practices RAG_ENABLED=true

1. Navigate to: Settings → Webhooks
2. Add webhook URL: https://your-domain.com/webhook/gitlab
3. Secret Token: Use value from GITLAB_WEBHOOK_SECRET
4. Trigger events:
   • ✅ Merge request events
   • ✅ Comments
5. Enable SSL verification

1. Create a new Dify application
2. Upload frontend best practices knowledge base:
   • React/Vue/Angular guidelines
   • Accessibility standards (WCAG)
   • Performance optimization patterns
   • Security best practices
   • CSS/HTML conventions
3. Enable RAG retrieval mode
4. Copy API key to .env

Automated Setup:

# Upload knowledge base to Dify automatically npm run setup:knowledge-base

This script will:

• Read all markdown files from knowledge-base/ directory
• Create a dataset in Dify
• Upload and index all documents
• Configure retrieval settings

Manual Setup:

Alternatively, you can manually upload files from knowledge-base/ to Dify:

1. Go to Dify Console → Datasets
2. Create new dataset: "frontend-best-practices"
3. Upload markdown files from knowledge-base/frontend/
4. Wait for indexing to complete
5. Link dataset to your Dify app

Local Fallback:

The application includes a local RAG fallback that reads knowledge base files directly from the filesystem when Dify is unavailable. This ensures the bot continues working even without external RAG service.

# Development npm run dev # Production npm start # With PM2 npm run start:pm2

Enhance code review quality by providing repository-specific context. Create a .aicodereview file with information about your codebase:

# Copy example template cp examples/.aicodereview.example .aicodereview # Edit with your repository details nano .aicodereview

The context file can include:

• Repository Type: monorepo, polyrepo, or library
• Project Structure: Description of directories and their purposes
• Code Style Guidelines: Project-specific coding standards
• Review Focus Areas: What to prioritize (performance, security, accessibility)
• Skip Patterns: Additional files/folders to exclude from review
• Custom Instructions: Special requirements for your team

Using context in manual reviews:

# Trigger review with repository context gitlab-ai-review review -p 12345 -m 42 --context ./.aicodereview # Short form gitlab-ai-review review -p 12345 -m 42 -c ./.aicodereview

The AI will use this context to provide more relevant, project-specific feedback.

Upload best practices to Dify RAG:

npm run setup:knowledge-base

This uploads all files from knowledge-base/ directory to Dify, enabling context-aware code reviews with:

• ✅ React best practices
• ✅ Vue.js patterns
• ✅ Angular 18+ guidelines
• ✅ Accessibility (WCAG)
• ✅ Performance optimization
• ✅ Security best practices

Note: The bot works even without Dify RAG - it falls back to reading local files from knowledge-base/ directory.

curl -X POST https://your-domain.com/api/review \ -H "Content-Type: application/json" \ -H "Authorization: Bearer YOUR_API_KEY" \ -d '{  "projectId": "12345",  "mrIid": 42  }'

curl https://your-domain.com/health

gitlab-ai-review/ ├── src/ │ ├── config/ │ │ ├── index.js # Configuration loader │ │ └── constants.js # Application constants │ ├── services/ │ │ ├── gitlab.service.js # GitLab API client │ │ ├── dify.service.js # Dify API client │ │ ├── review.service.js # Review orchestration │ │ └── rag.service.js # RAG knowledge base │ ├── middleware/ │ │ ├── auth.middleware.js # Authentication │ │ ├── validation.middleware.js # Request validation │ │ ├── rateLimit.middleware.js # Rate limiting │ │ └── error.middleware.js # Error handling │ ├── controllers/ │ │ ├── webhook.controller.js # Webhook handler │ │ └── review.controller.js # Manual review │ ├── utils/ │ │ ├── logger.js # Winston logger │ │ ├── errors.js # Custom errors │ │ ├── helpers.js # Helper functions │ │ └── prompt-loader.js # Prompt template loader │ ├── routes/ │ │ ├── index.js # API routes │ │ └── webhook.routes.js # Webhook routes │ └── app.js # Express app ├── tests/ │ ├── setup.js # Test configuration │ ├── unit/ │ │ ├── services/ │ │ └── utils/ │ ├── integration/ │ │ └── review.test.js │ └── fixtures/ ├── docs/ │ ├── CLI.md # CLI documentation │ ├── COST_ANALYSIS.md # Cost analysis │ └── FAQ.md # Frequently asked questions ├── scripts/ │ ├── setup.sh # Setup script │ └── deploy.sh # Deployment script ├── knowledge-base/ # Knowledge base & prompts │ ├── prompts/ # AI prompt templates │ │ ├── actionable-review.md # Default review prompt │ │ ├── critical-only-review.md # Critical issues only │ │ └── detailed-review.md # Comprehensive review │ ├── frontend/ # Frontend best practices (RAG) │ │ ├── react-best-practices.md │ │ ├── vue-best-practices.md │ │ ├── accessibility.md │ │ ├── performance.md │ │ └── security.md │ ├── backend/ # Backend best practices (RAG) │ │ ├── api-design.md │ │ └── security.md │ └── README.md # Knowledge base documentation ├── logs/ # Application logs (gitignored) ├── .github/ │ └── workflows/ │ ├── ci.yml # CI pipeline │ └── deploy.yml # Deployment pipeline ├── .husky/ # Git hooks │ └── pre-commit ├── Dockerfile ├── docker-compose.yml ├── ecosystem.config.js # PM2 configuration ├── .env.example ├── .eslintrc.js ├── .prettierrc ├── .gitignore ├── .dockerignore ├── jest.config.js ├── package.json ├── examples/ │ ├── .aicodereview.example # Full context template │ └── .aicodereview.minimal # Minimal context template └── README.md 

The .aicodereview file (optional, placed in project root) provides project-specific information to the AI:

# Repository Context ## Repository Type: monorepo ## Project Structure - apps/api - NestJS BFF - apps/ui - React UI client ## Code Style Guidelines - Airbnb style guide - Strict TypeScript mode ## Review Focus Areas - Performance critical for UI - Security critical for API ## Skip Patterns - apps/ui-e2e/\*\* (E2E tests)

Use with CLI:

gitlab-ai-review review -p <project> -m <mr> --context ./.aicodereview

# Linting npm run lint npm run lint:fix # Formatting npm run format # Type checking npm run typecheck

# Run all tests npm test # Watch mode npm run test:watch # Coverage npm run test:coverage # E2E tests npm run test:e2e

# Debug mode npm run debug # VS Code launch configuration provided

# Build image docker build -t gitlab-ai-review:latest . # Run container docker run -d \ --name gitlab-review-bot \ -p 3000:3000 \ --env-file .env \ gitlab-ai-review:latest

docker-compose up -d

pm2 start ecosystem.config.js pm2 save pm2 startup

Deploy using Docker to various cloud platforms:

• AWS ECS/Fargate: Use docker image with ECS task definition
• Google Cloud Run: Deploy from container registry
• Azure Container Instances: Use az container create
• Kubernetes: Apply deployment manifests with docker image

# Unit tests npm run test:unit # Integration tests npm run test:integration # Load testing npm run test:load

The application exposes metrics at /metrics:

• Request duration
• Review success/failure rate
• Dify API latency
• Memory usage
• Active connections

Recommended monitoring stack:

• Prometheus for metrics
• Grafana for visualization
• Sentry for error tracking

• API key authentication
• Webhook signature verification
• Rate limiting
• Input validation
• Dependency scanning (Snyk/Dependabot)
• OWASP security headers

MIT License - see LICENSE

Contributions are welcome! Please:

1. Fork the repository
2. Create a feature branch
3. Make your changes with tests
4. Submit a pull request

• 📧 Email: support@yourcompany.com
• 💬 Slack: #gitlab-ai-review
• 🐛 Issues: GitHub Issues

• Dify AI team for the excellent RAG platform
• GitLab community for API documentation
• OpenAI for GPT models
• All contributors and early adopters

See CHANGELOG.md for version history and updates.

• Dify - LLM app development platform
• GitLab - DevOps platform
• Review Bot - GitLab's official bot

Made with ❤️ for better code reviews