AI-Integrated Code Pipeline

This is an AI powered action designed to enhance code review and analyse file changes within a pull request by leveraging Gemini Nano to evaluate code quality, identify potential risks, suggest improvements, and generate tests automatically. It integrates seamlessly with your GitHub workflows to provide insightful feedback directly within your pull requests.

---

Usage

Prerequisites

• A GitHub repository
• GitHub Actions enabled
• An account on Bunjy AI

Step 1: Generate API Credentials

1. Visit Bunjy AI
2. Log in with your GitHub account
3. On the homepage sidebar click and navigate to "Access Tokens"
4. Generate a new API Key and Base app url by clicking the generate button
5. Copy your unique API_KEY and BASE_APP_URL

Step 2: Configure Repository Secrets

In your GitHub repository:

1. Go to "Settings"
2. Select "Secrets and variables"
3. Click "New repository secret"
4. Add the two secrets:
   • Name: BASE_APP_URL
     • Value: Your unique base URL from Bunjy AI
   • Name: API_KEY
     • Value: Your generated API key
5. NOTE: The action DOES NOT needs a personal TOKEN from github, if not provided it defaults to the one provided by github actions automatically.

I will soon find a way to use the action without the BASE_APP_URL specified, making it more flexible and easier to configure.

Step 3: Create Workflow File

Create a .github/workflows/ai.yml file in your repository, use this template to set up:

name: AI Code Review

on:
  pull_request:
    branches:
      '*'

jobs:
  ai:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Bunjy AI Code Review              
        uses: mohswell/mintify@v2.1
        with:
          BASE_APP_URL: ${{ secrets.BASE_APP_URL }}
          API_KEY: ${{ secrets.API_KEY }}
          GENERATE_TESTS: 'true' # Optional: defaults to 'false' if not provided

To enable automatic generation of unit tests for changed files, set the GENERATE_TESTS input to 'true' in your workflow file:

To cover fork branches and also to disable automatic test generation, you can modify the GENERATE_TESTS value to 'false'. use this template to set up the workflow:

name: AI Code Review

on:
  pull_request_target:
    types: [opened, synchronize, reopened]

jobs:
  ai:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Bunjy AI Code Review              
        uses: mohswell/mintify@v2.1
        with:
          BASE_APP_URL: ${{ secrets.BASE_APP_URL }}
          API_KEY: ${{ secrets.API_KEY }}
          GENERATE_TESTS: 'false'

How It Works: Technical Architecture

1. Authentication and Integration

• Users setup the action in their workflows.
• The action requires a BASE_APP_URL and API_KEY defined in your respository secrets.
• You'll visit the dashboard page at Bunjy AI to generate an API_KEY and also copy the BASE_APP_URL.
• Users can log in using GitHub OAuth to authorize my core service.
• The application connects directly to GitHub repositories via the action workflow.
• Securely retrieves repository and pull request data.
• Encrypts the repository data and sends it to the web server for processing/decrypting and then sends it to Gemini AI.
• Displays AI reviews for your code changes directly within the PR opened, or in the dahboard homepage.

2. AI Analysis Process

1. When a pull request is opened, the app:
   • Captures code changes
   • Analyzes commit messages
   • Processes repository context
   • Runs AI models (Gemini Nano) to evaluate:
     • Code quality
     • Potential risks
     • Improvement suggestions
     • Reviewer recommendations

3. Data Flow

flowchart LR
    A[GitHub PR] --> B[Core API Service]
    B --> C[Gemini AI]
    C --> D[Analysis Results]
    D --> E[Supabase DB]
    E --> F[Web Interface]
    F --> G{User Feedback}
    G -->|Review/Update| B
    D -->|Detailed Insights| H[Reporting Tools]
    H -->|Export/Share| I[External Platforms]

Loading

---

🛠️ What Does This App Do?

This application provides intelligent insights into GitHub pull requests by analyzing:

• Code changes, commit messages, and statistics.
• AI-driven suggestions for PR quality improvements.
• PR risk assessments based on AI analysis.
• Reviewer suggestions and code quality highlights.
• Insights displayed directly in GitHub Pull Requests
• Comprehensive analysis available in Bunjy AI dashboard

Apps and Packages

• web: The main user-facing Next.js application for viewing and managing pull request analyses.
• api: The backend core service powered by Prisma and Nest.js for interacting with the database and integrating AI services like Gemini.
• workflows/scripts: This repository includes various shell scripts to automate different tasks:

1. Analyzes files in the repository to gather insights and metrics.
2. Sets up the necessary environment variables required for the application.
3. Sends metadata to the server for further processing and analysis.
4. Processes comments in pull requests to provide feedback and suggestions.

---

Installation

Clone the repository and install dependencies:

git clone https://github.com/mohswell/mintify.git
cd mintify
npm install

npm install turbo --save-dev

---

Environment Variables

Configure environment variables in .env, .env.local, or .env.template for different environments.

Root .env Example:

DATABASE_URL=""
DIRECT_URL=""

For the Frontend go to web and create a .env.local Example:

NEXT_PUBLIC_SUPABASE_URL=""
NEXT_PUBLIC_SUPABASE_ANON_KEY=""
SUPABASE_SERVICE_ROLE_KEY=""
NEXTAUTH_URL="http://localhost:3000"
NEXTAUTH_SECRET=""
NEXT_PUBLIC_AUTH_REDIRECT_URL="http://localhost:3000"
GITHUB_CLIENT_ID=""
GITHUB_SECRET=""
NEXT_PUBLIC_API_BASE_URL="http://localhost:8000/api/v1"

For the Backend go to api and create a .env Example:

GEMINI_API_KEY=""
GEMINI_PRO_MODEL="gemini-pro"
GEMINI_PRO_VISION_MODEL="gemini-pro-vision"
PORT=8000
JWT_SECRET=""
DATABASE_URL=""

---

Scripts

Here are the key npm scripts for managing the monorepo:

Command	Description
turbo run build	Builds all apps and packages.
rubo run dev	Starts the development servers for all apps.
turbo run lint	Runs ESLint across all apps and packages.
turbo run format	Formats code using Prettier.
turbo run start	Starts the whole application in production mode.

---

Development

To start the development environment for all apps and packages:

turbo run dev

To build all apps and packages for production:

turbo run build

---

License

The repository is licensed under GNU GENERAL PUBLIC LICENSE, ensuring proper attribution and restricted commercial use. See the LICENSE file for details.

---

☕ Buy Me a Coffee

If you find this project helpful and would like to support its development, consider buying me a coffee:

Your support is greatly appreciated!

---

Contributing

Contributions are welcome! Please fork the repo and create a pull request with your changes and tag me in to review your features.

Future Roadmap

• Enhanced AI models
• More granular code analysis
• Machine learning improvements
• Extended language support

---

Contact

If you have questions or suggestions, feel free to open an issue or contact me directly at Muhammad.Said.

---