Local AI Reviewer 🔒

🔒 Privacy-first AI code reviewer that never sends your code anywhere!

Analyze your code for security vulnerabilities, performance issues, and bugs using powerful local AI models. 100% offline, 100% private.

Automatically review your code for security vulnerabilities, performance issues, bugs, and maintainability problems using local AI models. No data leaves your machine!

Features

✅ Local AI Models - Uses Ollama for complete privacy
✅ Multi-language Support - JavaScript, TypeScript, Python, Java, C++, Go, Rust, PHP
✅ Comprehensive Analysis - Security, performance, bugs, style, maintainability
✅ Multiple Output Formats - Console, JSON, Markdown
✅ Configurable - Customize patterns, severity levels, and review types
✅ CLI & Programmatic - Use as command-line tool or library

Installation

npm install -g local-ai-reviewer

Prerequisites

Install Ollama for local AI models:

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Start Ollama service
ollama serve

# Pull a code model (recommended)
ollama pull codellama:7b

Quick Start

1. Initialize configuration:

ai-review init

2. Review your code:

ai-review review

3. Review specific directory:

ai-review review ./src

Configuration

The ai-review.json file created by init:

{
  "model": "codellama:7b",
  "temperature": 0.1,
  "maxTokens": 2000,
  "includePatterns": ["**/*.{js,ts,jsx,tsx,py,java}"],
  "excludePatterns": ["**/node_modules/**", "**/dist/**"],
  "reviewTypes": ["security", "performance", "bugs", "maintainability"],
  "outputFormat": "console",
  "severity": "medium"
}

CLI Commands

# Initialize config file
ai-review init

# Review current directory
ai-review review

# Review with custom options
ai-review review ./src --model codellama:13b --output markdown

# Save report to file
ai-review review --save report.md

# List recommended models
ai-review models

Review Types

• Security - SQL injection, XSS, insecure patterns
• Performance - Inefficient algorithms, memory leaks
• Bugs - Logic errors, null pointer exceptions
• Style - Code formatting, naming conventions
• Maintainability - Complex functions, code smells
• Complexity - Cyclomatic complexity, nested loops

Output Formats

Console (Default)

ai-review review

JSON

ai-review review --output json

Markdown

ai-review review --output markdown --save report.md

Recommended Models

| Model | Size | Description | |-------|------|-------------| | codellama:7b | 3.8GB | Fast, good for basic review | | codellama:13b | 7.3GB | Better accuracy, slower | | deepseek-coder:6.7b | 3.8GB | Excellent for code analysis | | starcoder:7b | 4.1GB | Good for multiple languages |

Programmatic Usage

import { AICodeReviewer, ReviewConfig } from 'local-ai-reviewer';

const config: ReviewConfig = {
  model: 'codellama:7b',
  temperature: 0.1,
  maxTokens: 2000,
  includePatterns: ['**/*.ts'],
  excludePatterns: ['**/*.test.ts'],
  reviewTypes: ['security', 'bugs'],
  outputFormat: 'json',
  severity: 'medium'
};

const reviewer = new AICodeReviewer(config);
const result = await reviewer.review('./src');
const report = await reviewer.generateReport(result);

console.log(report);

Example Output

🤖 AI Code Review Report
==================================================

📊 Summary:
  Files Reviewed: 15
  Total Issues: 8
  Execution Time: 45.2s

🎯 Issues by Severity:
  HIGH: 2
  MEDIUM: 4
  LOW: 2

🔍 Detailed Issues:
--------------------------------------------------

1. Potential SQL Injection Vulnerability
   File: src/database.ts (Line 23)
   Type: security | Severity: HIGH

   Description:
   Direct string concatenation in SQL query without parameterization
   
   💡 Suggestion:
   Use parameterized queries or prepared statements

Privacy & Security

• ✅ 100% Local - All processing happens on your machine
• ✅ No Data Transmission - Code never leaves your environment
• ✅ Open Source - Full transparency
• ✅ Offline Capable - Works without internet connection

Supported Languages

• JavaScript/TypeScript
• Python
• Java
• C/C++
• Go
• Rust
• PHP
• Ruby
• C#
• Swift
• Kotlin

Contributing

1. Fork the repository
2. Create your feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

MIT

Support

• 📖 Documentation
• 🐛 Issues
• 💬 Discussions

Made with ❤️ for developers who value privacy