@tvmt/commit-reporter

AI-powered code review automation with Google Gemini & OpenAI. Analyze commits and generate comprehensive review reports.

Features

• 🤖 Three Specialized AI Agents:

  • Code Review Agent: Analyzes code quality, best practices, and potential bugs
  • Security Agent: Scans for security vulnerabilities (OWASP Top 10)
  • Performance Agent: Detects performance anti-patterns with AST analysis

• 📊 Comprehensive Reports: Generate both Markdown and HTML reports

• 🔄 Sequential Pipeline: Agents run in sequence, each building on previous findings

• 🎯 TypeScript/JavaScript Focus: Filters and analyzes only TS/JS files

• ⚙️ Flexible Configuration: Environment variables and CLI options

• 🔌 Multiple LLM Providers: Google Gemini & OpenAI support

🚀 Quick Start

Installation

# Global installation (recommended)
npm install -g @tvmt/commit-reporter

# Or use with npx (no installation needed)
npx @tvmt/commit-reporter review

Setup

1. Create .env file in your project:

# Choose your LLM provider
LLM_PROVIDER=google-genai

# Add your API key
GEMINI_API_KEY=your-api-key-here
# Or for OpenAI
# OPENAI_API_KEY=sk-...

2. Run review:

commit-reporter review

Installation from Source

# Clone the repository
git clone https://github.com/tranvuminhtriet/code-reviewer.git
cd code-reviewer

# Install dependencies
npm install

# Build
npm run build

# Link globally
npm link

Quick Start

Review a Git Commit

# Review the latest commit
npm run review

# Review a specific commit
npm run review -- --commit abc123

# Review HEAD~1
npm run review -- --commit HEAD~1

Review from a Diff File

# Use the example diff
npm run review -- --file examples/sample-diff.txt

# Use your own diff file
npm run review -- --file path/to/your.diff

Configuration

Environment Variables (.env)

# LLM Provider
LLM_PROVIDER=openai
OPENAI_API_KEY=your-key-here
OPENAI_MODEL=gpt-4-turbo-preview

# Output
OUTPUT_DIR=./reports
OUTPUT_FORMATS=markdown,html

# Agents (set to false to disable)
ENABLE_CODE_REVIEW=true
ENABLE_SECURITY=true
ENABLE_PERFORMANCE=true

CLI Options

code-review review [options]

Options:
  -c, --commit <hash>      Git commit hash (default: "HEAD")
  -f, --file <path>        Diff file path
  --api-key <key>          OpenAI API key (overrides env)
  --model <name>           Model name
  --output <dir>           Output directory
  --format <formats>       Output formats: markdown,html
  --no-code-review         Disable code review agent
  --no-security            Disable security agent
  --no-performance         Disable performance agent
  -h, --help               Display help

Examples

Basic Usage

# Review latest commit with default settings
npm run review

# Output:
# ✓ Parsed 2 file(s): 2 files changed, 30 insertions(+), 5 deletions(-)
# 🤖 Starting AI code review...
# Running Code Review Agent...
# ✓ Code Review: 5 findings
# Running Security Agent...
# ✓ Security: 3 findings
# Running Performance Agent...
# ✓ Performance: 2 findings
# ✅ Review complete!

Custom Configuration

# Only run security and performance agents
npm run review -- --no-code-review

# Use a different model
npm run review -- --model gpt-4

# Output only markdown
npm run review -- --format markdown

# Custom output directory
npm run review -- --output ./my-reports

Review Specific Files

# Create a diff of specific files
git diff HEAD~1 -- src/api/*.ts > my-changes.diff

# Review the diff
npm run review -- --file my-changes.diff

Report Output

Reports are generated in the specified output directory (default: ./reports):

reports/
├── code-review-2024-02-05T10-30-00.md
└── code-review-2024-02-05T10-30-00.html

Report Structure

1. Summary: Total findings by severity and agent
2. Token Usage: LLM token consumption
3. Code Review Findings: Code quality issues
4. Security Findings: Security vulnerabilities
5. Performance Findings: Performance issues

Each finding includes:

• Severity (Critical, High, Medium, Low)
• Category
• Description
• File and line number
• Suggestion for fix

Architecture

┌─────────────────┐
│  Git Diff       │
│  Parser         │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Code Review    │
│  Agent          │
└────────┬────────┘
         │ (findings)
         ▼
┌─────────────────┐
│  Security       │
│  Agent          │
│  + ESLint       │
└────────┬────────┘
         │ (findings)
         ▼
┌─────────────────┐
│  Performance    │
│  Agent          │
│  + AST Analysis │
└────────┬────────┘
         │ (findings)
         ▼
┌─────────────────┐
│  Report         │
│  Generator      │
│  (MD + HTML)    │
└─────────────────┘

Development

Build

npm run build

Run in Development Mode

npm run dev review -- --file examples/sample-diff.txt

Project Structure

src/
├── agents/           # AI agents
│   ├── base-agent.ts
│   ├── code-review-agent.ts
│   ├── security-agent.ts
│   └── performance-agent.ts
├── llm/              # LLM provider abstraction
│   ├── types.ts
│   ├── provider.ts
│   └── openai-provider.ts
├── parsers/          # Git diff parser
│   ├── types.ts
│   └── git-diff-parser.ts
├── reporters/        # Report generators
│   ├── types.ts
│   ├── base-reporter.ts
│   ├── markdown-reporter.ts
│   └── html-reporter.ts
├── pipeline/         # Orchestration
│   ├── types.ts
│   └── executor.ts
├── config/           # Configuration
│   └── default.ts
└── cli.ts            # CLI entry point

Future Enhancements

Phase 2: Additional LLM Providers

# Anthropic Claude
npm run review -- --provider anthropic --api-key sk-ant-...

# Local Ollama
npm run review -- --provider ollama --model codellama

Phase 3: CI/CD Integration

# .github/workflows/code-review.yml
name: AI Code Review
on: [pull_request]
jobs:
  review:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
      - run: npm install
      - run: npm run review -- --commit ${{ github.event.pull_request.head.sha }}
      - uses: actions/upload-artifact@v3
        with:
          name: code-review-report
          path: reports/

Phase 4: Migration to LangGraph

For more complex workflows, the architecture can be migrated to LangGraph:

import { StateGraph } from '@langchain/langgraph';

const workflow = new StateGraph({...});
workflow
  .addNode('codeReview', codeReviewAgent)
  .addNode('security', securityAgent)
  .addNode('performance', performanceAgent)
  .addEdge('codeReview', 'security')
  .addEdge('codeReview', 'performance');

Troubleshooting

"API key is required" Error

Make sure you have set OPENAI_API_KEY in your .env file or pass it via CLI:

npm run review -- --api-key sk-...

"No TypeScript/JavaScript files found"

The tool only analyzes .ts, .tsx, .js, and .jsx files. Check that your diff contains these file types.

ESLint Errors

If you encounter ESLint parsing errors, the security agent will skip those files and continue with others.

License

MIT

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.