mbot – A GitHub PR Code Reviewer

mbot is an AI-powered GitHub PR reviewer built with FastAPI. It leverages Gemini 1.5 Pro to analyze pull requests and provide smart suggestions for code improvements.

Features

• Automated GitHub PR Code Review → Uses GitHub Webhooks to trigger automated code reviews for pull requests, analyzing changes and providing insights.
• Built with FastAPI ,Uvicorn & uv → High-performance backend leveraging FastAPI for speed, Uvicorn as the ASGI server, and uv for dependency management.
• Secure GitHub Authorization → Authenticates using a GitHub App with secure API access via private keys.
• Dockerized for Easy Deployment → Containerized with Docker, ensuring portability and seamless deployment across environments.
• CI/CD with GitHub Actions → Automates building and pushing the latest Docker image to Docker Hub on every push to the main branch.
• Seamless Deployment on Google Cloud Run → Google Cloud Run automatically pulls the latest Docker image and deploys updates with zero downtime.

Installation

Clone the repository

  git clone https://github.com/rohanrjp/MBot.git

Run Locally with Docker

docker build -t mbot .
docker run -p 8080:8080 mbot

You can pull the latest Docker image from Docker Hub and run it using the following commands:

docker pull rjp1007/mbot
docker run -p 8080:8080 rjp1007/mbot:latest

Environment Variables

To run this project, you will need to add the following environment variables to your .env file

GEMINI_API_KEY -> API key for accessing Gemini AI to analyze and review code changes.

GITHUB_APP_ID -> The GitHub App ID used to authenticate as a GitHub App for PR review.

GITHUB_PRIVATE_KEY -> Private key for securely signing GitHub API requests as the app.

API Reference

Root endpoint

  GET /

Get Github webhook requests and post PR code review comments

  POST /api/pr_review

Tech Stack

Backend: FastAPI , Uvicorn
Deployment: Docker , Google Cloud Run , GitHub Actions
Development Tools: Ngrok , Httpx

Lessons Learned

What I learnt while building this project?

Building MBot provided valuable insights into GitHub webhooks, authorization, and automated CI/CD workflows. I deepened my understanding of FastAPI, uvicorn, and uv, optimizing API performance and request handling. Deploying the project using Docker and Google Cloud Run helped reinforce best practices for containerization and cloud deployment.

What challenges did you face, and how did you overcome them?

1. Webhook Handling & Authorization

Setting up GitHub webhooks and managing authentication with GitHub App ID and private keys required a precise configuration. I overcame this by thoroughly reading GitHub’s documentation and debugging using Ngrok to test webhook events locally.

2. Docker & Cloud Deployment Issues

Initially, I faced issues with exposed ports and Google Cloud Run failing to detect the correct service port. Fixing this required explicitly setting the PORT environment variable and ensuring that FastAPI was listening on 0.0.0.0:8080.

3. CI/CD Automation

Automating Docker image builds and deployments with GitHub Actions had some challenges with authentication (Docker Hub & Google Cloud). Using GitHub Secrets for secure credential management and refining the workflow YAML helped in streamlining deployments.

Each challenge strengthened my understanding of cloud-based API deployment, containerized applications, and CI/CD automation.

Screenshots

Roadmap

• Implement a feature that detects which lines of code were modified
• Expand AI capabilities to detect security vulnerabilities in PRs
• Explore fine-tuning a custom model for even better PR insights

Authors

• @rohanrjp

Feedback

If you have any feedback, please reach out to us at rohan1007rjp@gmail.com