PR Review Agent

AI-powered GitHub PR review platform that analyzes pull requests using Gemini, applies company coding standards, and securely stores review history in MongoDB.

Comment

Inspiration

Code reviews are one of the most important parts of software development, but they are also repetitive and time-consuming. Developers often spend significant time understanding pull requests, checking coding standards, and reviewing similar patterns repeatedly.

I wanted to build an AI-powered workspace that assists developers by generating structured code reviews while keeping every review searchable for future reference.

What it does

PR Review Agent is an AI-powered code review platform that helps developers review GitHub Pull Requests more efficiently.

Users can:

  • Sign up using Email/Password or Google OAuth
  • Import GitHub Pull Requests
  • Fetch live PR metadata directly from GitHub
  • Generate AI-powered review reports using Google's Gemini model
  • Include issue context and company-specific coding guidelines
  • Save every review permanently in MongoDB
  • Browse previous review history from a personal workspace
  • Optionally post generated reviews back to GitHub

The application provides structured feedback including:

  • Overall rating
  • Summary
  • Strengths
  • Potential issues
  • Suggestions
  • Code improvements

How I built it

The application follows a layered architecture:

Frontend

  • HTML
  • CSS
  • JavaScript

Backend

  • Node.js
  • Express

Database

  • MongoDB
  • Mongoose ODM

Authentication

  • Auth0
  • Google OAuth

AI

  • Google Gemini API

Version Control

  • GitHub API

Deployment Ready

  • Docker
  • Docker Compose

The project is organized using Controllers, Services, Repositories, Models, and Routes to keep the business logic clean and maintainable.

Challenges

Some of the biggest challenges included:

  • Designing a clean repository architecture
  • Migrating from file-based JSON storage to MongoDB without changing business logic
  • Managing authentication sessions securely
  • Integrating multiple external APIs (GitHub, Gemini, Auth0)
  • Building Docker support for consistent deployment
  • Designing prompts that generate useful and structured AI reviews

Accomplishments

  • Implemented complete authentication system
  • Added persistent MongoDB storage
  • Integrated live GitHub Pull Request fetching
  • Built reusable repository architecture
  • Added Docker and Docker Compose support
  • Created a modern responsive UI
  • Designed structured AI review generation

What I learned

This project significantly improved my understanding of:

  • Backend architecture
  • Repository pattern
  • MongoDB and Mongoose
  • OAuth authentication
  • Docker containerization
  • Prompt engineering
  • GitHub API integration
  • Production-ready application design

What's next

Future improvements include:

  • Team workspaces
  • Review collaboration
  • Inline code annotations
  • Multi-model AI support
  • CI/CD integration
  • Review analytics dashboard
  • Unit and integration tests
  • Cloud deployment

Built With

Share this project:

Updates

Submission history